Digital scholarship blog

25 posts categorized "Science"

23 October 2020

BL Labs Public Award Runner Up (Research) 2019 - Automated Labelling of People in Video Archives

Example people identified in TV news related programme clips
People 'automatically' identified in digital TV news related programme clips.

Guest blog post by Andrew Brown (PhD researcher),  Ernesto Coto (Research Software Engineer) and Andrew Zisserman (Professor) of the Visual Geometry Group, Department of Engineering Science, University of Oxford, and BL Labs Public Award Runner-up for Research, 2019. Posted on their behalf by Mahendra Mahey, Manager of BL Labs.

In this work, we automatically identify and label (tag) people in large video archives without the need for any manual annotation or supervision. The project was carried out with the British Library on a sample of 106 videos from their “Television and radio news” archive; a large collection of news programs from the last 10 years. This archive serves as an important and fascinating resource for researchers and the general public alike. However, the sheer scale of the data, coupled with a lack of relevant metadata, makes indexing, analysing and navigating this content an increasingly difficult task. Relying on human annotation is no longer feasible, and without an effective way to navigate these videos, this bank of knowledge is largely inaccessible.

As users, we are typically interested in human-centric queries such as:

  • “When did Jeremy Corbyn first appear in a Newsnight episode?” or
  • “Show me all of the times when Hugh Grant and Shirley Williams appeared together.

Currently this is nigh on impossible without trawling through hundreds of hours of content. 

We posed the following research question:

Is it possible to enable automatic person-search capabilities such as this in the archive, without the need for any manual supervision or labelling?

The answer is “yes”, and the method is described next.

Video Pre-Processing

The basic unit which enables person labelling in videos is the face-track; a group of consecutive face detections within a shot that correspond to the same identity. Face-tracks are extracted from all of the videos in the archive. The task of labelling the people in the videos is then to assign a label to each one of these extracted face-tracks. The video below gives an example of two face-tracks found in a scene.


Two face-tracks found in British Library digital news footage by Visual Geometry Group - University of Oxford.

Techniques at Our Disposal

The base technology used for this work is a state-of-the-art convolutional neural network (CNN), trained for facial recognition [1]. The CNN extracts feature-vectors (a list of numbers) from face images, which indicate the identity of the depicted person. To label a face-track, the distance between the feature-vector for the face-track, and the feature-vector for a face-image with known identity is computed. The face-track is labelled as depicting that identity if the distance is smaller than a certain threshold (i.e. they match). We also use a speaker recognition CNN [2] that works in the same way, except it labels speech segments from unknown identities using speech segments from known identities within the video.

Labelling the Face-Tracks

Our method for automatically labelling the people in the video archive is divided into three main stages:

(1) Our first labelling method uses what we term a “celebrity feature-vector bank”, which consists of names of people that are likely to appear in the videos, and their corresponding feature-vectors. The names are automatically sourced from IMDB cast lists for the programmes (the titles of the programmes are freely available in the meta-data). Face-images for each of the names are automatically downloaded from image-search engines. Incorrect face-images and people with no images of themselves on search engines are automatically removed at this stage. We compute the feature-vectors for each identity and add them to the bank alongside the names. The face-tracks from the video archives are then simply labelled by finding matches in the feature-vector bank.

Face-tracks from the video archives are labelled by finding matches in the feature-vector bank.
Face-tracks from the video archives are labelled by finding matches in the feature-vector bank. 

(2) Our second labelling method uses the idea that if a name is spoken, or found displayed in a scene, then that person is likely to be found within that scene. The task is then to automatically determine whether there is a correspondence or not. Text is automatically read from the news videos using Optical Character Recognition (OCR), and speech is automatically transcribed using Automatic Speech Recognition (ASR). Names are identified and they are searched for on image search engines. The top ranked images are downloaded and the feature-vectors are computed from the faces. If any are close enough to the feature-vectors from the face-tracks present in the scene, then that face-track is labelled with that name. The video below details this process for a written name.


Using text or spoken word and face recognition to identify a person in a news clip.

(3) For our third labelling method, we use speaker recognition to identify any non-labelled speaking people. We use the labels from the previous two stages to automatically acquire labelled speech segments from the corresponding labelled face-tracks. For each remaining non-labelled speaking person, we extract the speech feature-vector and compute the distance of it to the feature-vectors of the labelled speech segments. If one is close enough, then the non-labelled speech segment and corresponding face-track is assigned that name. This process manages to label speaking face-tracks with visually challenging faces, e.g. deep in shadow or at an extremely non-frontal pose.

Indexing and Searching Identities

The results of our work can be browsed via a web search engine of our own design. A search bar allows for users to specify the person or group of people that they would like to search for. People’s names are efficiently indexed so that the complete list of names can be filtered as the user types in the search bar. The search results are returned instantly with their associated metadata (programme name, data and time) and can be displayed in multiple ways. The video associated with each search result can be played, visualising the location and the name of all identified people in the video. See the video below for more details. This allows for the archive videos to be easily navigated using person-search, thus opening them up for use by the general public.


Archive videos easily navigated using person-search.

For examples of more of our Computer Vision research and open-source software, visit the Visual Geometry Group website.

This work was supported by the EPSRC Programme Grant Seebibyte EP/M013774/1

[1] Qiong Cao, Li Shen, Weidi Xie, Omkar M. Parkhi, and Andrew Zisserman. VGGFace2: A dataset for recognising faces across pose and age. In Proc. International Conference on Automatic Face & Gesture Recognition, 2018.

[2] Joon Son Chung, Arsha Nagrani and Andrew Zisserman. VoxCeleb2: Deep Speaker Recognition. INTERSPEECH, 2018

BL Labs Public Awards 2020

Inspired by this work that uses the British Library's digital archived news footage? Have you done something innovative using the British Library's digital collections and data? Why not consider entering your work for a BL Labs Public Award 2020 and win fame, glory and even a bit of money?

This year's public and staff awards 2020 are open for submission, the deadline for entry for both is Monday 30 November 2020.

Whilst we welcome projects on any use of our digital collections and data (especially in research, artistic, educational and community categories), we are particularly interested in entries in our public awards that have focused on anti-racist work, about the pandemic or that are using computational methods such as the use of Jupyter Notebooks.

19 October 2020

The 2020 British Library Labs Staff Award - Nominations Open!

Looking for entries now!

A set of 4 light bulbs presented next to each other, the third light bulb is switched on. The image is supposed to a metaphor to represent an 'idea'
Nominate an existing British Library staff member or a team that has done something exciting, innovative and cool with the British Library’s digital collections or data.

The 2020 British Library Labs Staff Award, now in its fifth year, gives recognition to current British Library staff who have created something brilliant using the Library’s digital collections or data.

Perhaps you know of a project that developed new forms of knowledge, or an activity that delivered commercial value to the library. Did the person or team create an artistic work that inspired, stimulated, amazed and provoked? Do you know of a project developed by the Library where quality learning experiences were generated using the Library’s digital content? 

You may nominate a current member of British Library staff, a team, or yourself (if you are a member of staff), for the Staff Award using this form.

The deadline for submission is NOON (GMT), Monday 30 November 2020.

Nominees will be highlighted on Tuesday 15 December 2020 at the online British Library Labs Annual Symposium where some (winners and runners-up) will also be asked to talk about their projects (everyone is welcome to attend, you just need to register).

You can see the projects submitted by members of staff and public for the awards in our online archive.

In 2019, last year's winner focused on the brilliant work of the Imaging Team for the 'Qatar Foundation Partnership Project Hack Days', which were sessions organised for the team to experiment with the Library's digital collections. 

The runner-up for the BL Labs Staff Award in 2019 was the Heritage Made Digital team and their social media campaign to promote the British Library's digital collections one language a week from letters 'A' to 'U' #AToUnknown).

In the public Awards, last year's winners (2019) drew attention to artisticresearchteaching & learning, and community activities that used our data and / or digital collections.

British Library Labs is a project within the Digital Scholarship department at the British Library that supports and inspires the use of the Library's digital collections and data in exciting and innovative ways. It was previously funded by the Andrew W. Mellon Foundation and is now solely funded by the British Library.

If you have any questions, please contact us at labs@bl.uk.

11 September 2020

BL Labs Public Awards 2020: enter before NOON GMT Monday 30 November 2020! REMINDER

The sixth BL Labs Public Awards 2020 formally recognises outstanding and innovative work that has been carried out using the British Library’s data and / or digital collections by researchers, artists, entrepreneurs, educators, students and the general public.

The closing date for entering the Public Awards is NOON GMT on Monday 30 November 2020 and you can submit your entry any time up to then.

Please help us spread the word! We want to encourage any one interested to submit over the next few months, who knows, you could even win fame and glory, priceless! We really hope to have another year of fantastic projects to showcase at our annual online awards symposium on the 15 December 2020 (which is open for registration too), inspired by our digital collections and data!

This year, BL Labs is commending work in four key areas that have used or been inspired by our digital collections and data:

  • Research - A project or activity that shows the development of new knowledge, research methods, or tools.
  • Artistic - An artistic or creative endeavour that inspires, stimulates, amazes and provokes.
  • Educational - Quality learning experiences created for learners of any age and ability that use the Library's digital content.
  • Community - Work that has been created by an individual or group in a community.

What kind of projects are we looking for this year?

Whilst we are really happy for you to submit your work on any subject that uses our digital collections, in this significant year, we are particularly interested in entries that may have a focus on anti-racist work or projects about lock down / global pandemic. We are also curious and keen to have submissions that have used Jupyter Notebooks to carry out computational work on our digital collections and data.

After the submission deadline has passed, entries will be shortlisted and selected entrants will be notified via email by midnight on Friday 4th December 2020. 

A prize of £150 in British Library online vouchers will be awarded to the winner and £50 in the same format to the runner up in each Awards category at the Symposium. Of course if you enter, it will be at least a chance to showcase your work to a wide audience and in the past this has often resulted in major collaborations.

The talent of the BL Labs Awards winners and runners up over the last five years has led to the production of remarkable and varied collection of innovative projects described in our 'Digital Projects Archive'. In 2019, the Awards commended work in four main categories – Research, Artistic, Community and Educational:

BL_Labs_Winners_2019-smallBL  Labs Award Winners for 2019
(Top-Left) Full-Text search of Early Music Prints Online (F-TEMPO) - Research, (Top-Right) Emerging Formats: Discovering and Collecting Contemporary British Interactive Fiction - Artistic
(Bottom-Left) John Faucit Saville and the theatres of the East Midlands Circuit - Community commendation
(Bottom-Right) The Other Voice (Learning and Teaching)

For further detailed information, please visit BL Labs Public Awards 2020, or contact us at labs@bl.uk if you have a specific query.

Posted by Mahendra Mahey, Manager of British Library Labs.

20 May 2020

Bringing Metadata & Full-text Together

This is a guest post by enthusiastic data and metadata nerd Andy Jackson (@anjacks0n), Technical Lead for the UK Web Archive.

In Searching eTheses for the openVirus project we put together a basic system for searching theses. This only used the information from the PDFs themselves, which meant the results looked like this:

openVirus EThOS search results screen
openVirus EThOS search results screen

The basics are working fine, but the document titles are largely meaningless, the last-modified dates are clearly suspect (26 theses in the year 1600?!), and the facets aren’t terribly useful.

The EThOS metadata has much richer information that the EThOS team has collected and verified over the years. This includes:

  • Title
  • Author
  • DOI, ISNI, ORCID
  • Institution
  • Date
  • Supervisor(s)
  • Funder(s)
  • Dewey Decimal Classification
  • EThOS Service URL
  • Repository (‘Landing Page’) URL

So, the question is, how do we integrate these two sets of data into a single system?

Linking on URLs

The EThOS team supplied the PDF download URLs for each record, but we need a common identifer to merge these two datasets. Fortunately, both datasets contain the EThOS Service URL, which looks like this:

https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.755301

This (or just the uk.bl.ethos.755301 part) can be used as the ‘key’ for the merge, leaving us with one data set that contains the download URLs alongside all the other fields. We can then process the text from each PDF, and look up the URL in this metadata dataset, and merge the two together in the same way.

Except… it doesn’t work.

The web is a messy place: those PDF URLs may have been direct downloads in the past, but now many of them are no longer simple links, but chains of redirects. As an example, this original download URL:

http://repository.royalholloway.ac.uk/items/bf7a78df-c538-4bff-a28d-983a91cf0634/1/10090181.pdf

Now redirects (HTTP 301 Moved Permanently) to the HTTPS version:

https://repository.royalholloway.ac.uk/items/bf7a78df-c538-4bff-a28d-983a91cf0634/1/10090181.pdf

Which then redirects (HTTP 302 Found) to the actual PDF file:

https://repository.royalholloway.ac.uk/file/bf7a78df-c538-4bff-a28d-983a91cf0634/1/10090181.pdf

So, to bring this all together, we have to trace these links between the EThOS records and the actual PDF documents.

Re-tracing Our Steps

While the crawler we built to download these PDFs worked well enough, it isn’t quite a sophisticated as our main crawler, which is based on Heritrix 3. In particular, Heritrix offers details crawl logs that can be used to trace crawler activity. This functionality would be fairly easy to add to Scrapy, but that’s not been done yet. So, another approach is needed.

To trace the crawl, we need to be able to look up URLs and then analyse what happened. In particular, for every starting URL (a.k.a. seed) we want to check if it was a redirect and if so, follow that URL to see where it leads.

We already use content (CDX) indexes to allow us to look up URLs when accessing content. In particular, we use OutbackCDX as the index, and then the pywb playback system to retrieve and access the records and see what happened. So one option is to spin up a separate playback system and query that to work out where the links go.

However, as we only want to trace redirects, we can do something a little simpler. We can use the OutbackCDX service to look up what we got for each URL, and use the same warcio library that pywb uses to read the WARC record and find any redirects. The same process can then be repeated with the resulting URL, until all the chains of redirects have been followed.

This leaves us with a large list, linking every URL we crawled back to the original PDF URL. This can then be used to link each item to the corresponding EThOS record.

This large look-up table allowed the full-text and metadata to be combined. It was then imported into a new Solr index that replaced the original service, augmenting the records with the new metadata.

Updating the Interface

The new fields are accessible via the same API as before – see this simple search as an example.

The next step was to update the UI to take advantage of these fields. This was relatively simple, as it mostly involved exchanging one field name for another (e.g. from last_modified_year to year_i), and adding a few links to take advantage of the fact we now have access to the URLs to the EThOS records and the landing pages.

The result can be seen at:

EThOS Faceted Search Prototype

The Results

This new service provides a much better interface to the collection, and really demonstrates the benefits of combining machine-generated and manually curated metadata.

New openVirus EThOS search results interface
New improved openVirus EThOS search results interface

There are still some issues with the source data that need to be resolved at some point. In particular, there are now only 88,082 records, which indicates that some gaps and mismatches emerged during the process of merging these records together.

But it’s good enough for now.

The next question is: how do we integrate this into the openVirus workflow? 

 

14 May 2020

Searching eTheses for the openVirus project

This is a guest post by Andy Jackson (@anjacks0n), Technical Lead for the UK Web Archive and enthusiastic data-miner.

Introduction

The COVID-19 outbreak is an unprecedented global crisis that has prompted an unprecedented global response. I’ve been particularly interested in how academic scholars and publishers have responded:

It’s impressive how much has been done in such a short time! But I also saw one comment that really stuck with me:

“Our digital libraries and archives may hold crucial clues and content about how to help with the #covid19 outbreak: particularly this is the case with scientific literature. Now is the time for institutional bravery around access!”
– @melissaterras

Clearly, academic scholars and publishers are already collaborating. What could digital libraries and archives do to help?

Scale, Audience & Scope

Almost all the efforts I’ve seen so far are focused on helping scientists working on the COVID-19 response to find information from publications that are directly related to coronavirus epidemics. The outbreak is much bigger than this. In terms of scope, it’s not just about understanding the coronavirus itself. The outbreak raises many broader questions, like:

  • What types of personal protective equipment are appropriate for different medical procedures?
  • How effective are the different kinds of masks when it comes to protecting others?
  • What coping strategies have proven useful for people in isolation?

(These are just the examples I’ve personally seen requests for. There will be more.)

Similarly, the audience is much wider than the scientists working directly on the COVID-19 response. From medical professions wanting to know more about protective equipment, to journalists looking for context and counter-arguments.

As a technologist working at the British Library, I felt like there must be some way I could help this situation. Some way to help a wider audience dig out any potentially relevant material we might hold?

The openVirus Project

While looking out for inspiration, I found Peter Murray-Rust’s openVirus project. Peter is a vocal supporter of open source and open data, and had launched an ambitious attempt to aggregate information relating to viruses and epidemics from scholarly publications.

In contrast to the other efforts I’d seen, Peter wanted to focus on novel data-mining methods, and on pulling in less well-known sources of information. This dual focus on text analysis and on opening up underutilised resources appealed to me. And I already had a particular resource in mind…

EThOS

Of course, the British Library has a very wide range of holdings, but as an ex-academic scientist I’ve always had a soft spot for EThOS, which provides electronic access to UK theses.

Through the web interface, users can search the metadata and abstracts of over half a million theses. Furthermore, to support data mining and analysis, the EThOS metadata has been published as a dataset. This dataset includes links to institutional repository pages for many of the theses.

Although doctoral theses are not generally considered to be as important as journal articles, they are a rich and underused source of information, capable of carrying much more context and commentary than a brief article[1].

The Idea

Having identified EThOS as source of information, the idea was to see if I could use our existing UK Web Archive tools to collect and index the full-text of these theses, build a simple faceted search interface, and perform some basic data-mining operations. If that worked, it would allow relevant theses to be discovered and passed to the openVirus tools for more sophisticated analysis.

Preparing the data sources

The links in the EThOS dataset point to the HTML landing-page for each theses, rather than to the full text itself. To get to the text, the best approach would be to write a crawler to find the PDFs. However, it would take a while to create something that could cope with the variety of ways the landing pages tend to be formatted. For machines, it’s not always easy to find the link to the actual theses!

However, many of the universities involved have given the EThOS team permission to download a copy of their theses for safe-keeping. The URLs of the full-text files are only used once (to collect each thesis shortly after publication), but have nevertheless been kept in the EThOS system since then. These URLs are considered transient (i.e. likely to ‘rot’ over time) and come with no guarantees of longer-term availability (unlike the landing pages), so are not included in the main EThOS dataset. Nevertheless, the EThOS team were able to give me the list of PDF URLs, making it easier to get started quickly.

This is far from ideal: we will miss theses that have been moved to new URLs, and from universities that do not take part (which, notably, includes Oxford and Cambridge). This skew would be avoided if we were to use the landing-page URLs provided for all UK digital theses to crawl the PDFs. But we need to move quickly.

So, while keeping these caveats in mind, the first task was to crawl the URLs and see if the PDFs were still there…

Collecting the PDFs

A simple Scrapy crawler was created, one that could read the PDF URLs and download them without overloading the host repositories. The crawler itself does nothing with them, but by running behind warcprox the web requests and responses (including the PDFs) can be captured in the standardised Web ARChive (WARC) format.

For 35 hours, the crawler attempted to download the 130,330 PDF URLs. Quite a lot of URLs had already changed, but 111,793 documents were successfully downloaded. Of these, 104,746 were PDFs.

All the requests and responses generated by the crawler were captured in 1,433 WARCs each around 1GB in size, totalling around 1.5TB of data.

Processing the WARCs

We already have tools for handling WARCs, so the task was to re-use them and see what we get. As this collection is mostly PDFs, Apache Tika and PDFBox are doing most of the work, but the webarchive-discovery wrapper helps run them at scale and add in additional metadata.

The WARCs were transferred to our internal Hadoop cluster, and in just over an hour the text and associated metadata were available as about 5GB of compressed JSON Lines.

A Legal Aside

Before proceeding, there’s legal problem that we need to address. Despite being freely-available over the open web, the rights and licenses under which these documents are being made available can be extremely varied and complex.

There’s no problem gathering the content and using it for data mining. The problem is that there are limitations on what we can redistribute without permission: we can’t redistribute the original PDFs, or any close approximation.

However, collections of facts about the PDFs are fine.

But for the other openVirus tools to do their work, we need to be able to find out what each thesis are about. So how can we make this work?

One answer is to generate statistical summaries of the contents of the documents. For example, we can break the text of each document up into individual words, and count how often each word occurs. These word frequencies are a no substitute for the real text, but are redistributable and suitable for answering simple queries.

These simple queries can be used to narrow down the overall dataset, picking out a relevant subset. Once the list of documents of interest is down to a manageable size, an individual researcher can download the original documents themselves, from the original hosts[2]. As the researcher now has local copies, they can run their own tools over them, including the openVirus tools.

Word Frequencies

second, simpler Hadoop job was created, post-processing the raw text and replacing it with the word frequency data. This produced 6GB of uncompressed JSON Lines data, which could then be loaded into an instance of the Apache Solr search tool [3].

While Solr provides a user interface, it’s not really suitable for general users, nor is it entirely safe to expose to the World Wide Web. To mitigate this, the index was built on a virtual server well away from any production systems, and wrapped with a web server configured in a way that should prevent problems.

The API this provides (see the Solr documentation for details) enables us to find which theses include which terms. Here are some example queries:

This is fine for programmatic access, but with a little extra wrapping we can make it more useful to more people.

APIs & Notebooks

For example, I was able to create live API documentation and a simple user interface using Google’s Colaboratory:

Using the openVirus EThOS API

Google Colaboratory is a proprietary platform, but those notebooks can be exported as more standard Jupyter Notebooks. See here for an example.

Faceted Search

Having carefully exposed the API to the open web, I was also able to take an existing browser-based faceted search interface and modify to suite our use case:

EThOS Faceted Search Prototype

Best of all, this is running on the Glitch collaborative coding platform, so you can go look at the source code and remix it yourself, if you like:

EThOS Faceted Search Prototype – Glitch project

Limitations

The main limitation of using word-frequencies instead of full-text is that phrase search is broken. Searching for face AND mask will work as expected, but searching for “face mask” doesn’t.

Another problem is that the EThOS metadata has not been integrated with the raw text search. This would give us a much richer experience, like accurate publication years and more helpful facets[4].

In terms of user interface, the faceted search UI above is very basic, but for the openVirus project the API is likely to be of more use in the short term.

Next Steps

To make the search more usable, the next logical step is to attempt to integrate the full-text search with the EThOS metadata.

Then, if the results look good, we can start to work out how to feed the results into the workflow of the openVirus tool suite.

 


1. Even things like negative results, which are informative but can be difficult to publish in article form. ↩︎

2. This is similar data sharing pattern used by Twitter researchers. See, for example, the DocNow Catalogue. ↩︎

3. We use Apache Solr a lot so this was the simplest choice for us. ↩︎

4. Note that since writing this post, this limitation has been rectified. ↩︎

 

03 October 2019

BL Labs Symposium (2019): Book your place for Mon 11-Nov-2019

Posted by Mahendra Mahey, Manager of BL Labs

The BL Labs team are pleased to announce that the seventh annual British Library Labs Symposium will be held on Monday 11 November 2019, from 9:30 - 17:00* (see note below) in the British Library Knowledge Centre, St Pancras. The event is FREE, and you must book a ticket in advance to reserve your place. Last year's event was the largest we have ever held, so please don't miss out and book early!

*Please note, that directly after the Symposium, we have teamed up with an interactive/immersive theatre company called 'Uninvited Guests' for a specially organised early evening event for Symposium attendees (the full cost is £13 with some concessions available). Read more at the bottom of this posting!

The Symposium showcases innovative and inspiring projects which have used the British Library’s digital content. Last year's Award winner's drew attention to artistic, research, teaching & learning, and commercial activities that used our digital collections.

The annual event provides a platform for the development of ideas and projects, facilitating collaboration, networking and debate in the Digital Scholarship field as well as being a focus on the creative reuse of the British Library's and other organisations' digital collections and data in many other sectors. Read what groups of Master's Library and Information Science students from City University London (#CityLIS) said about the Symposium last year.

We are very proud to announce that this year's keynote will be delivered by scientist Armand Leroi, Professor of Evolutionary Biology at Imperial College, London.

Armand Leroi
Professor Armand Leroi from Imperial College
will be giving the keynote at this year's BL Labs Symposium (2019)

Professor Armand Leroi is an author, broadcaster and evolutionary biologist.

He has written and presented several documentary series on Channel 4 and BBC Four. His latest documentary was The Secret Science of Pop for BBC Four (2017) presenting the results of the analysis of over 17,000 western pop music from 1960 to 2010 from the US Bill Board top 100 charts together with colleagues from Queen Mary University, with further work published by through the Royal Society. Armand has a special interest in how we can apply techniques from evolutionary biology to ask important questions about culture, humanities and what is unique about us as humans.

Previously, Armand presented Human Mutants, a three-part documentary series about human deformity for Channel 4 and as an award winning book, Mutants: On Genetic Variety and Human Body. He also wrote and presented a two part series What Makes Us Human also for Channel 4. On BBC Four Armand presented the documentaries What Darwin Didn't Know and Aristotle's Lagoon also releasing the book, The Lagoon: How Aristotle Invented Science looking at Aristotle's impact on Science as we know it today.

Armands' keynote will reflect on his interest and experience in applying techniques he has used over many years from evolutionary biology such as bioinformatics, data-mining and machine learning to ask meaningful 'big' questions about culture, humanities and what makes us human.

The title of his talk will be 'The New Science of Culture'. Armand will follow in the footsteps of previous prestigious BL Labs keynote speakers: Dan Pett (2018); Josie Fraser (2017); Melissa Terras (2016); David De Roure and George Oates (2015); Tim Hitchcock (2014); Bill Thompson and Andrew Prescott in 2013.

The symposium will be introduced by the British Library's new Chief Librarian Liz Jolly. The day will include an update and exciting news from Mahendra Mahey (BL Labs Manager at the British Library) about the work of BL Labs highlighting innovative collaborations BL Labs has been working on including how it is working with Labs around the world to share experiences and knowledge, lessons learned . There will be news from the Digital Scholarship team about the exciting projects they have been working on such as Living with Machines and other initiatives together with a special insight from the British Library’s Digital Preservation team into how they attempt to preserve our digital collections and data for future generations.

Throughout the day, there will be several announcements and presentations showcasing work from nominated projects for the BL Labs Awards 2019, which were recognised last year for work that used the British Library’s digital content in Artistic, Research, Educational and commercial activities.

There will also be a chance to find out who has been nominated and recognised for the British Library Staff Award 2019 which highlights the work of an outstanding individual (or team) at the British Library who has worked creatively and originally with the British Library's digital collections and data (nominations close midday 5 November 2019).

As is our tradition, the Symposium will have plenty of opportunities for networking throughout the day, culminating in a reception for delegates and British Library staff to mingle and chat over a drink and nibbles.

Finally, we have teamed up with the interactive/immersive theatre company 'Uninvited Guests' who will give a specially organised performance for BL Labs Symposium attendees, directly after the symposium. This participatory performance will take the audience on a journey through a world that is on the cusp of a technological disaster. Our period of history could vanish forever from human memory because digital information will be wiped out for good. How can we leave a trace of our existence to those born later? Don't miss out on a chance to book on this unique event at 5pm specially organised to coincide with the end of the BL Labs Symposium. For more information, and for booking (spaces are limited), please visit here (the full cost is £13 with some concessions available). Please note, if you are unfortunate in not being able to join the 5pm show, there will be another performance at 1945 the same evening (book here for that one).

So don't forget to book your place for the Symposium today as we predict it will be another full house again and we don't want you to miss out.

We look forward to seeing new faces and meeting old friends again!

For any further information, please contact labs@bl.uk

10 June 2019

Collaborative Digital Scholarship in Action: A Case Study in Designing Impactful Student Learning Partnerships

The Arts and Sciences (BASc) department at University College London has been at the forefront of pioneering a renascence of liberal arts and sciences degrees in the UK. As part of its Core modules offering, students select an interdisciplinary elective in Year 2 of their academic programme – from a range of modules specially designed for the department by University College London academics and researchers.

When creating my own module – Information Through the Ages (BASC0033) – as part of this elective set, I was keen to ensure that the student learning experience was both supported and developed in tandem with professional practices and standards, knowing that enabling students to progress their skills developed on the module beyond the module’s own assignments would aid them not only in their own unique academic degree programmes but also provide substantial evidence to future employers of their employability and skills base. Partnering with the British Library, therefore, in designing a data science and data curation project as part of the module’s core assignments, seemed to me to provide an excellent opportunity to enable both a research-based educative framework for students as well as a valuable chance for them to engage in a real-world collaboration, as providing students with external industry partners to collaborate with can contribute an important fillip to their motivation and the learning experience overall – by seeing their assessed work move beyond the confines of the academy to have an impact out in the wider world.

Through discussions with my British Library co-collaborators, Mahendra Mahey and Stella Wisdom, we alighted on the Microsoft Books/BL 19th Century collection dataset as providing excellent potential for student groups to work with for their data curation projects. With its 60,000 public domain volumes, associated metadata and 1 million+ extracted images, it presented as exciting, undiscovered territory across which our student groups might roam and rove, with the results of their work having the potential to benefit future British Library researchers.

Structuring the group project around wrangling a subset of this data: discovering, researching, cleaning and refining it, with the output from each group a curated version of the original dataset we therefore felt presented a number of significant benefits. Students were enabled to explore and develop technical skills such as data curation, software knowledge, archival research, report writing, project development and collaborative working practices, alongside experiencing a real world, digital scholarship learning experience – with the outcomes in turn supporting the British Library’s Digital Scholarship remit regards enabling innovative research based on the British Library digital collections.

Students observed that “working with the data did give me more practical insight to the field of work involved with digitisation work, and it was an enriching experience”, including how they “appreciated how involved and hands-on the projects were, as this is something that I particularly enjoy”. Data curation training was provided on site at the British Library, with the session focused on the use of OpenRefine, “a powerful tool for working with messy data: cleaning it; transforming it from one format into another; and extending it with web services and external data.”[1] Student feedback also told us that we could have provided further software training, and more guided dataset exploration/navigation resources, with groups keen to learn more nuanced data curation techniques – something we will aim to respond to in future iterations of the module – but overall, as one student succinctly noted, “I had no idea of the digitalization process and I learned a lot about data science. The training was very useful and I acquired new skills about data cleaning.”

Overall, we had five student groups wrangling the BL 19th Century collection, producing final data subsets in the following areas: Christian and Christian-related texts; Queens of Britain 1510-1946; female authors, 1800-1900 (here's a heatmap this student group produced of the spread of published titles by female authors in the 19th century); Shakespearean works, other author’s adaptations on those works, and any commentary on Shakespeare or his writing; and travel-related books.

In particular, it was excellent to see students fully engaging with the research process around their chosen data subset – exploring its cultural and institutional contexts, as well as navigating metadata/data schemas, requirements and standards.

For example, the Christian texts group considered the issue of different languages as part of their data subset of texts, following this up with textual content analysis to enable accurate record querying and selection. In their project report they noted that “[u]sing our dataset and visualisations as aids, we hope that researchers studying the Bible and Christianity can discover insights into the geographical and temporal spread of Christian-related texts. Furthermore, we hope that they can also glean new information regarding the people behind the translations of Bibles as well as those who wrote about Christianity.”

Similarly, the student group focused on travel-related texts discussed in their team project summary that “[t]he particular value of this curated dataset is that future researchers may be able to use it in the analysis of international points of view. In these works, many cities and nations are being written about from an outside perspective. This perspective is one that can be valuable in understanding historical relations and frames of reference between groups around the world: for instance, the work “Travels in France and Italy, in 1817 and 1818”, published in New York, likely provides an American perspective of Europe, while “Four Months in Persia, and a Visit to Trans-Caspia”, published in London, might detail an extended visit of a European in Persia, both revealing unique perspectives about different groups of people. A comparable work, that may have utilized or benefitted from such a collection, is Hahner’s (1998) “Women Through Women’s Eyes:Latin American Women in Nineteenth Century Travel Accounts.” In it, Hahner explores nineteenth century literature written to unearth the perspectives on Latin American women, specifically noting that the primarily European author’s writings should be understood in the context of their Eurocentric view, entrenched in “patriarchy” and “colonialism” (Hahner, 1998:21). Authors and researchers with a similar intent may use [our] curated British Library dataset comparably – that is, to locate such works.”

Data visualisation by travel books group
Data visualisation by travel books group
Data visualisation by travel books group
Data visualisation by travel books group

Over the ten weeks of the module, alongside their group data curation projects, students covered lecture topics as varied as Is a Star a Document?, "Truthiness" and Truth in a Post-Truth World, Organising Information: Classification, Taxonomies and Beyond!, and Information & Power; worked on an individual archival GIF project which drew on an institutional archival collection to create (and publish on social media) an animated GIF; and spent time in classroom discussions considering questions such as What happens when information is used for dis-informing or mis-informing purposes?; How do the technologies available to us in the 21st century potentially impact on the (data) collection process and its outputs and outcomes?; How might ideas about collections and collecting be transformed in a digital context?; What exactly do we mean by the concepts of Data and Information?; How we choose to classify or group something first requires we have a series of "rules" or instructions which determine the grouping process – but who decides on what the rules are and how might such decisions in fact influence our very understandings of the information the system is supposedly designed to facilitate access to? These dialogues were all situated within the context of both "traditional" collections systems and atypical sites of information storage and collection, with the module aiming to enable students to gain an in-depth knowledge, understanding and critical appreciation of the concept of information, from historical antecedents to digital scientific and cultural heritage forms, in the context of libraries, archives, galleries and museums (including alternative, atypical and emergent sources), and how technological, social, cultural and other changes fundamentally affect our concept of “information.”

“I think this module was particularly helpful in making me look at things in an interdisciplinary light”, one student observed in module evaluation feedback, with others going on to note that “I think the different formats of work we had to do was engaging and made the coursework much more interesting than just papers or just a project … the collaboration with the British Library deeply enriched the experience by providing a direct and visible outlet for any energies expended on the module. It made the material seem more applicable and the coursework more enjoyable … I loved that this module offered different ways of assessment. Having papers, projects, presentations, and creative multimedia work made this course engaging.”

Situating the module’s assessments within such contexts I hope encouraged students to understand the critical, interdisciplinary focus of the field of information studies, in particular the use of information in the context of empire-making and consolidation, and how histories of information, knowledge and power intersect. Combined with a collaborative, interdisciplinary curriculum design approach, which encouraged and supported students to gain technical abilities and navigate teamwork practices, we hope this module can point some useful ways forward in creating and developing engaging learning experiences, which have real world impact.

This blog post is by Sara Wingate-Gray (UCL Senior Teaching Fellow & BASC0033 module leader), Mahendra Mahey (BL Labs Manager) and Stella Wisdom (BL Digital Curator for Contemporary British Collections).

25 January 2019

BL Labs 2018 Artistic Award Winner: 'Another Intelligence Sings'

This guest blog is by the winners of the BL Labs Artistic Award for 2018, Robert Walker, Rose Leahy and Amanda Baum, for 'Another Intelligence Sings'.AI Sings 1

When the natural world is recorded, it is quantised for the human ear, to wavelengths within our perception and timeframes within our conception. Yet the machine learning algorithm sits outside the human sensorium, outside the human lifespan. An algorithm is agnostic to the source, the intention and the timescale of data. By feeding it audio samples of lava and larvae, geological tensions and fleeting courtship, the seismic and the somatic, the many voices of life are woven into a song no one lifespan or life form could sing.

Another Intelligence Sings ( AI Sings ) is an immersive audio-tactile installation inviting you to experience the sounds of our biological world as recounted through an AI. Through the application of neural networks to field recordings from the British Library sound archive a nonhuman reading of the data emerges. Presenting an alternative composition of Earth’s songs, AI Sings explores an expanded view of what might be perceived as intelligent.

The breadth of the British Library Wildlife and Environmental Sounds archive enabled us to take a cross section of the natural world from primordial physical phenomena to the great beasts of the savannas to the songbirds of the British countryside. The final soundscape is created from using two different neural networks, Wavenet and Nsynth. We trained Wavenet, Google’s most advanced human speech synthesis neural network, on many hours of field recordings, including those from the British Library archives.

Nsynth is an augmented version of Wavenet that was built and trained by Magenta, Google AI’s creative lab. Nsynth creates sounds that are not a simple crossfade or blend but something genuinely new based on the perceived formal musical qualities of the two source sounds. This was used to create mixtures between specific audio samples, for example, sea lion meets mosquito, leopard meets horse, and mealworm meets ocean.

Click here to play a 4 minute clip of the sound from the installation.

AI Sings 2
Through this use of the technology, AI Sings reorients the algorithm’s focus, away from the human expression of individual thought and towards an amalgam of geological and biological processes. The experience aims to enable humans to meditate on the myriad intelligences around and beyond us and expand our view of what might be perceived as intelligent. This feeds into our ongoing body of shared work, which raises questions about the use of artificial intelligence in society. Previously, we have used a neural network to find linguistic patterns not perceivable to human reading to mediate our collectively written piece Weaving Worlds (2016). In AI Sings we continue this thread of asking which perspectives an AI can bring that human perception cannot.

AI Sings 3

AI Sings takes digital archive content and makes it into a tactile, sensuous, and playful experience. By making the archive material an experiential encounter, we were able to encourage listeners to enter into a world where they could be immersed and engaged in the data. Soft, tactile materials such as hair and foam invited people to enter into and interact with the work. In particular, we found that the playful nature of the materials in the piece meant that children were keen to experience the work, and listen to the soundscape, thereby extending the audience of the archive material to one it may not usually reach.

By addressing the need for experiential, visceral and poetic encounters with AI, Another Intelligence Sings goes beyond the conceptual and engages people in the technology which is so rapidly transforming society. We hope this work shows how the creative application of AI opens up new possibilities in the field of archivology, from being a tool of categorisation to becoming a means of expanding the cultural role of the library in the future.

The piece premiered at the V&A Digital Design Weekend 2018 on 22nd of September as part of London Design Festival, where it was exhibited to over 22,000 visitors. Following the weekend we were invited by Open Cell, London’s newly opened bioart- and biodesign studio and exhibition space, to be showcased on their site.

More about the project can be found on our websites:

www.baumleahy.com + www.irr.co + www.amandabaum.com + www.roseleahy.com

Watch the AI Sings team receiving their award and talking about their project on our YouTube channel (clip runs from 8.18):

 

Find out more about Digital Scholarship and BL Labs. If you have a project which uses British Library digital content in innovative and interesting ways, consider applying for an award this year! The 2019 BL Labs Symposium will take place on Monday 11 November at the British Library.

Digital scholarship blog recent posts

Archives

Tags

Other British Library blogs