British LIbrary Labs Presentation to Bodelian Interns

Editor's Notes

• #2: 46 seconds Hello everyone, <click>my name’s Mahendra Mahey. I have been working at the British Library for over a year and I manage a Digital Lab which is responsible for opening up the British Library’s digital content and data and getting researchers to use it for their research, my colleague Ben O’Steen is responsible for providing a technical lead for the project and supporting researchers who want to use our content and data by providing advice and practical technical support and help. I will talk give an introduction to British Library Labs a brief overview of the surprising things that happened when we worked with one of our digital collections.
• #3: 1 min 34 The Labs project’s is trying to ‘open up’ the digital content and data (such as digitised books, images and maps) <click> the British Library has and encourage researchers, particularly in the Arts and Humanities area to ‘Experiment with our digital collections’ <click> to use it for their scholarly work and ideas and consider research questions that only become possible to ask when you are providing access to large amounts of digital content and data. Our website is available at labs.bl.uk. The project is kindly funded by the Andrew Mellon Foundation. We work closely with our colleagues in the Digital Research Team (Stella Wisdom, Nora McGregor, James Baker, Aquiles Alencar Brayner and Rossizta Atanassova) colleagues in the Digital Scholarship Department, around the Library and researchers and other colleagues around the UK and world wide.
• #4: 1min 58 seconds <click>We adopt a ‘Data Driven’ approach to encourage scholars to do research and development with and across British Library digital collections and data, i.e. here is our stuff… come with some interesting research ideas and questions of what to do with it and we will work with you and try and support you <click>A researcher will have domain knowledge about a subject but not necessarily the technical skills to be provide answers to the questions they are posing. A developer / computer scientist may have the technical skills to deal with large amounts of data but may not have enough domain knowledge to ask the right questions. Sometimes these people exist in the same body and sometimes they need to work together. We want them to come up with an idea and engage with Labs through various mechanisms <click>such as competitions, events (Hack events, workshops, ideas labs and projects). The kinds of methods that are being considered will be looked at in a bit more detail on the next slide Libraries. Through this process the Library is learning how better to support scholars and to build on existing processes or create new ones, as well as make <click>tools (e.g. APIs etc.) and services. The <click>case studies are some of the outputs we hope to create that will help other research libraries around the world wanting to build Labs for their digital content, <click> others include open software and publications.
• #5: 180 seconds So what kinds digital research methods are these digital scholars using especially in the area of Digital Humanities? <click>For example, searching for items based on and time location can reveal very interesting patterns, e.g. when and where works were published. For example one researcher is looking at the evidence of copy and paste in newspapers in the 19th Century which was a common practice back then. Knowing where and when items might include text from a source can reveal patterns of how the text travelled over time. <click>Geotagging objects, putting them in space can add new dimensions to the kinds of research questions we might want to ask. <click>Corpus analysis is the analysis of text in Language and Text mining is about finding patterns in text through computational analysis for example number crunching (a lot of it based on counting words). <click>Tasks that require humans fall under the area of Crowdsourcing and Human Computation for example e.g. recaptcha is used as a way to get users to contribute to better text from scanned books but typing in the word they think they see in order to get access to a service. Amazon’s Mechanical Turk is a form of human computation, where tasks are outsourced to humans that computers would find very hard to do. <click>Annotation involves augmenting an item with additional information, usually text, but not necessarily, e.g. highlighting an area, a drawing etc. <click>Natural Language processing is used in the analysis of speech, for example. <click>Similarly transcribing can be the conversion of speech into text through human or computing power to then be used for further analysis. <click>Providing Application Programming Interfaces or APIs to data can be very powerful ways to access datasets, and can be used by software developers to build software applications on top of them, the term the API economy is now very popular and is fuelling the development of lots of apps for mobile phones <click>Many researchers want to see the patterns that are emerging in large amounts of data and are now using a number of very powerful tools to visualise data to see patterns. What is clear is that digital research methods are much more that searching for an individual item in a catalogue and Libraries, publishers, service and content providers have to change to support that.
• #6: 1 min 20 seconds A major part of Labs activity is to run an annual competition. We encourage researchers to look at our data, talk to us, and more importantly to talk to each other and submit ideas and project plans of what they could do in a 4-6 month residency at the British Library. This ‘residency model ‘enables researchers to get access to pretty much all the digital content they require without any license restrictions and we get to engage with them deeply to learn about what they want to do and importantly what we need to learn as a library to support digital scholars better. We work in an agile way with the researchers. <click>We worked with two researchers who won last year’s competition, <click>Pieter Francois and Dan Norton over a <click>4 month period to work on their research questions and ideas. Let’s now look briefly at their ideas and what was achieved.
• #7: 65 seconds Dr Dan Norton was researcher at the University of Dundee and artist in residence at Hangar, Centre for Art and Research, Barcelona. His idea was “Mixing the Library: The Disc Jockey and the Digital Collection” which brought a DJ’s approach to interacting with multi-format digital collections <click>. Dan’s interactive approach helps build aesthetic, experimental, or logical links between resources. This project focused on ideas around creating a prototype<click> and what would be the basic building blocks needed to create a simple demonstrator <click>. Dan is now building on the work he did at Labs and is the resident researcher and artist at the <click>Living Labs: Library of the Future in Barcelona, where he will be working with software developers to continue his work.
• #8: Let’s look at a brief video of about Pieter and his entry. 1 min 30
• #9: 1 min 31 seconds We have announced our winners for our 2014 competition in June and are currently working with them, the results will be showcased at a large event on the 3 November 2014 in London. They are: Anna Gerber and Desmond Schmidt from Queensland University in Australia. In order to make old printed books and manuscripts accessible to a Web audience, it is essential to display the page image / facsimile of the original document next to its transcription. This allows the user to comment on the text, and to read it clearly, but because original documents are often hard to read, or have different line-breaks than text on a computer screen, it is easy to get lost trying to match up words in the document with words in the transcription. To overcome this, the team are developing semi-automatic methods to generate links that highlight corresponding parts of the page image and the text. Bob Nicholson, lecturer at Edge Hill University. The ‘Victorian Meme Machine’ [VMM] will create an extensive database of Victorian jokes that will be available for use by other scholars. Let’s have a look at a video explaining his entry.
• #10: 1 min 30 seconds
• #11: 75 seconds The work of Labs is really about a number of stories, stories about digital collections and about researchers wanting to ask fascinating research questions about them. Let’s now tell you a story about one collection and the intended and unintended consequences of working with it. The Library digitised 68,000 17th to 19th century books from our collections a few years ago (around 2.7 % of the physical total in that period. We also captured 22 million individual page images, along with full text scans of these images all of which contain untold quantity of useful data such as names of people, places, historical events, dates. So the question became then, what next? What can 68,000 books tell us?
• #12: 60 seconds As the books were scanned for text, this had a fortunate ‘side effect’ the software not only tries to detect the text on the page <click>but also where the images might be. <click>There had already been some interest as part of the competition, Matt Prior attended one of our hack events and when examining our book data and was very interested in the images from the books.
• #13: 50 seconds Ben O’Steen, Labs Technical lead, wrote an algorithm to begin to extract all the images that were picked up in the scanning process from the 68,000 books. He did this initially because he had an interest in face recognition algorithms, and wanted to know if algorithms that had been trained on passport photos could be used to detect faces in the images in this collection, many of which were illustrations<click>. He found that it was good at detecting <click>female faces but not men, as men tended to have beards and glasses!
• #14: 130 seconds Labs and other members of Digital Research team decided to post an image every hour on a tumblr blog <click>This was the first image that was published. The service was christened with the somewhat controversial name <click>Mechanical Curator (we like to be a little controversial) and said that it was a ‘she’. Our newest staff member churned away day and night posting an image every hour. It posted previously unseen illustrations taken almost at random. The Mechanical curator uses algorithms to chose other similar images based on a number of algorithms it has at hand, for example <click>e.g. published date, <click>slantyness, <click>bubblyness on the x or <click>y axis, or a<click> ‘new train of thought’ if it get’s bored. However, little was known about the actual image, apart from the analytic work of the Mechanical curator. Meanwhile the algorithm that Ben had written to snip the images from the OCR scans was still churning away, how many were there going to be? The Mechanical Curator could publish them every hour, but was there somewhere we could put them all for people to browse when they wanted. Importantly if we did put them somewhere, could we get people to help us add descriptions to the individual images making them infinitely more discoverable.
• #15: 65 seconds How many images do you think we found? <click>Over 1 million images were then put onto <click>Flickr commons. Why? Because each image would have a <click>URL. Each image had some <click>metadata, i.e. the book and page number it came from. However the image itself didn’t have any metadata, i.e. was it a picture of a dog? By releasing the images onto Flickr, we could begin to see if people might start adding tags to the images. <click>Flickr has a well known and used API which developers and researchers could start using to build applications on top of or for examining large amounts of them at the same time. People have already started putting images into sets as you can see from the picture, portraits and ships are very popular.
• #16: 35 seconds The site was launched on Friday, December 13, 2013. How many views do you think there have been? <click>There are have been a staggering 176,026,465 views of the images since December 13th, 2013, figures correct July 28. So Friday the 13th wasn’t so unlucky for us after all!
• #17: 50 seconds We are working with <click>Metadata Games to develop branded British Library games rapidly increase the amount metadata that is being added, 3 games have been launched, portrait tag was launched on the 18th of June<click>, e.g. we have lots of ships, maps and portraits to name a few. There is also an Arts and Humanities Research Council Big Data project called Lost Visions, <click>which will be answering detailed research questions about the images and doing some tagging too.
• #18: 35 seconds There are risks in this of course, surely lurking in the 1 million are images are sordid and of an offensive nature, especially given some of the views that were around at that time? <click>In the end we decided to not interfere, and take any issues as they may arise on a case-by-case basis. To date we’ve had very few.
• #19: 50 seconds Here is the anatomy of a Flickr record, importantly we have created links to many of the Library’s services <click>some of this lovely traffic is going back to the Library and hopefully generating more interest in our services, from downloading a pdf of the book to purchasing a high res scan of the image. <click>Tags are added from the original book record, including the approximate page number the image came from<click>users of Flickr can add their own tags, and I have mentioned they have already started doing it.
• #20: 18 seconds There has been considerable news coverage about the million images released on Flickr commons. <click>The Independent, <click>Wired magazine, <click>The Guardian, <click>Popular Science and the <click>Mail online to name a very few.
• #21: 15 seconds There have been several creative uses of the images and can be found at the website above. The creation of a skateboard which you can buy for $64. A colouring book! http://www.playingbythebook.net/2014/03/18/barbapapas-new-house-a-book-so-good-im-featuring-it-for-a-second-time/ A youtube video showing off the pictures. https://www.youtube.com/watch?v=uiS1cx38rKk&feature=youtu.be
• #22: Burning man is a surprising use of the images we put onto Flickr. Once a year, tens of thousands of participants gather in Nevada's Black Rock Desert to create Black Rock City, dedicated to community, art, self-expression, and self-reliance. They depart one week later, having left no trace whatsoever. This year it is taking place between August 25 to September 1, Nevada, USA, the show ends by burning an effigy of wooden man! <click> David Normal is using images from your Flickr Commons collection and is working on a set of collages called "Crossroads of Curiosity".  The finished images will be paintings based on these collages that will be presented in full colour as ' lightboxes at this year's Burning Man Festival.  These works are part of the festival's "Caravansary" theme for 2014, and will be presented around the base of the effigy of the Burning Man in the heart of the festival.