Digital scholarship blog

Enabling innovative research with British Library digital collections

Introduction

Tracking exciting developments at the intersection of libraries, scholarship and technology. Read more

20 April 2022

Importing images into Zooniverse with a IIIF manifest: introducing an experimental feature

Digital Curator Dr Mia Ridge shares news from a collaboration between the British Library and Zooniverse that means you can more easily create crowdsourcing projects with cultural heritage collections. There's a related blog post on Zooniverse, Fun with IIIF.

IIIF manifests - text files that tell software how to display images, sound or video files alongside metadata and other information about them - might not sound exciting, but by linking to them, you can view and annotate collections from around the world. The IIIF (International Image Interoperability Framework) standard makes images (or audio, video or 3D files) more re-usable - they can be displayed on another site alongside the original metadata and information provided by the source institution. If an institution updates a manifest - perhaps adding information from updated cataloguing or crowdsourcing - any sites that display that image automatically gets the updated metadata.

Playbill showing the title after other large text
Playbill showing the title after other large text

We've posted before about how we used IIIF manifests as the basis for our In the Spotlight crowdsourced tasks on LibCrowds.com. Playbills are great candidates for crowdsourcing because they are hard to transcribe automatically, and the layout and information present varies a lot. Using IIIF meant that we could access images of playbills directly from the British Library servers without needing server space and extra processing to make local copies. You didn't need technical knowledge to copy a manifest address and add a new volume of playbills to In the Spotlight. This worked well for a couple of years, but over time we'd found it difficult to maintain bespoke software for LibCrowds.

When we started looking for alternatives, the Zooniverse platform was an obvious option. Zooniverse hosts dozens of historical or cultural heritage projects, and hundreds of citizen science projects. It has millions of volunteers, and a 'project builder' that means anyone can create a crowdsourcing project - for free! We'd already started using Zooniverse for other Library crowdsourcing projects such as Living with Machines, which showed us how powerful the platform can be for reaching potential volunteers. 

But that experience also showed us how complicated the process of getting images and metadata onto Zooniverse could be. Using Zooniverse for volumes of playbills for In the Spotlight would require some specialist knowledge. We'd need to download images from our servers, resize them, generate a 'manifest' list of images and metadata, then upload it all to Zooniverse; and repeat that for each of the dozens of volumes of digitised playbills.

Fast forward to summer 2021, when we had the opportunity to put a small amount of funding into some development work by Zooniverse. I'd already collaborated with Sam Blickhan at Zooniverse on the Collective Wisdom project, so it was easy to drop her a line and ask if they had any plans or interest in supporting IIIF. It turns out they had, but hadn't had the resources or an interested organisation necessary before.

We came up with a brief outline of what the work needed to do, taking the ability to recreate some of the functionality of In the Spotlight on Zooniverse as a goal. Therefore, 'the ability to add subject sets via IIIF manifest links' was key. ('Subject set' is Zooniverse-speak for 'set of images or other media' that are the basis of crowdsourcing tasks.) And of course we wanted the ability to set up some crowdsourcing tasks with those items… The Zooniverse developer, Jim O'Donnell, shared his work in progress on GitHub, and I was very easily able to set up a test project and ask people to help create sample data for further testing. 

If you have a Zooniverse project and a IIIF address to hand, you can try out the import for yourself: add 'subject-sets/iiif?env=production' to your project builder URL. e.g. if your project is number #xxx then the URL to access the IIIF manifest import would be https://www.zooniverse.org/lab/xxx/subject-sets/iiif?env=production

Paste a manifest URL into the box. The platform parses the file to present a list of metadata fields, which you can flag as hidden or visible in the subject viewer (public task interface). When you're happy, you can click a button to upload the manifest as a new subject set (like a folder of items), and your images are imported. (Don't worry if it says '0 subjects).

 

Screenshot of manifest import screen
Screenshot of manifest import screen

You can try out our live task and help create real data for testing ingest processes at ​​https://frontend.preview.zooniverse.org/projects/bldigital/in-the-spotlight/classify

This is a very brief introduction, with more to come on managing data exports and IIIF annotations once you've set up, tested and launched a crowdsourced workflow (task). We'd love to hear from you - how might this be useful? What issues do you foresee? How might you want to expand or build on this functionality? Email [email protected] or tweet @mia_out @LibCrowds. You can also comment on GitHub https://github.com/zooniverse/Panoptes-Front-End/pull/6095 or https://github.com/zooniverse/iiif-annotations

Digital work in libraries is always collaborative, so I'd like to thank British Library colleagues in Finance, Procurement, Technology, Collection Metadata Services and various Collections departments; the Zooniverse volunteers who helped test our first task and of course the Zooniverse team, especially Sam, Jim and Chris for their work on this.

 

12 April 2022

Making British Library collections (even) more accessible

Daniel van Strien, Digital Curator, Living with Machines, writes:

The British Library’s digital scholarship department has made many digitised materials available to researchers. This includes a collection of digitised books created by the British Library in partnership with Microsoft. This is a collection of books that have been digitised and processed using Optical Character Recognition (OCR) software to make the text machine-readable. There is also a collection of books digitised in partnership with Google. 

Since being digitised, this collection of digitised books has been used for many different projects. This includes recent work to try and augment this dataset with genre metadata and a project using machine learning to tag images extracted from the books. The books have also served as training data for a historic language model.

This blog post will focus on two challenges of working with this dataset: size and documentation, and discuss how we’ve experimented with one potential approach to addressing these challenges. 

One of the challenges of working with this collection is its size. The OCR output is over 20GB. This poses some challenges for researchers and other interested users wanting to work with these collections. Projects like Living with Machines are one avenue in which the British Library seeks to develop new methods for working at scale. For an individual researcher, one of the possible barriers to working with a collection like this is the computational resources required to process it. 

Recently we have been experimenting with a Python library, datasets, to see if this can help make this collection easier to work with. The datasets library is part of the Hugging Face ecosystem. If you have been following developments in machine learning, you have probably heard of Hugging Face already. If not, Hugging Face is a delightfully named company focusing on developing open-source tools aimed at democratising machine learning. 

The datasets library is a tool aiming to make it easier for researchers to share and process large datasets for machine learning efficiently. Whilst this was the library’s original focus, there may also be other uses cases for which the datasets library may help make datasets held by the British Library more accessible. 

Some features of the datasets library:

  • Tools for efficiently processing large datasets 
  • Support for easily sharing datasets via a ‘dataset hub’ 
  • Support for documenting datasets hosted on the hub (more on this later). 

As a result of these and other features, we have recently worked on adding the British Library books dataset library to the Hugging Face hub. Making the dataset available via the datasets library has now made the dataset more accessible in a few different ways.

Firstly, it is now possible to download the dataset in two lines of Python code: 

Image of a line of code: "from datasets import load_dataset ds = load_dataset('blbooks', '1700_1799')"

We can also use the Hugging Face library to process large datasets. For example, we only want to include data with a high OCR confidence score (this partially helps filter out text with many OCR errors): 

Image of a line of code: "ds.filter(lambda example: example['mean_wc_ocr'] > 0.9)"

One of the particularly nice features here is that the library uses memory mapping to store the dataset under the hood. This means that you can process data that is larger than the RAM you have available on your machine. This can make the process of working with large datasets more accessible. We could also use this as a first step in processing data before getting back to more familiar tools like pandas. 

Image of a line of code: "dogs_data = ds['train'].filter(lamda example: "dog" in example['text'].lower()) df = dogs_data_to_pandas()

In a follow on blog post, we’ll dig into the technical details of datasets in some more detail. Whilst making the technical processing of datasets more accessible is one part of the puzzle, there are also non-technical challenges to making a dataset more usable. 

 

Documenting datasets 

One of the challenges of sharing large datasets is documenting the data effectively. Traditionally libraries have mainly focused on describing material at the ‘item level,’ i.e. documenting one dataset at a time. However, there is a difference between documenting one book and 100,000 books. There are no easy answers to this, but libraries could explore one possible avenue by using Datasheets. Timnit Gebru et al. proposed the idea of Datasheets in ‘Datasheets for Datasets’. A datasheet aims to provide a structured format for describing a dataset. This includes questions like how and why it was constructed, what the data consists of, and how it could potentially be used. Crucially, datasheets also encourage a discussion of the bias and limitations of a dataset. Whilst you can identify some of these limitations by working with the data, there is also a crucial amount of information known by curators of the data that might not be obvious to end-users of the data. Datasheets offer one possible way for libraries to begin more systematically commuting this information. 

The dataset hub adopts the practice of writing datasheets and encourages users of the hub to write a datasheet for their dataset. For the British library books, we have attempted to write one of these datacards. Whilst it is certainly not perfect, it hopefully begins to outline some of the challenges of this dataset and gives end-users a better sense of how they should approach a dataset. 

18 March 2022

Looking back at LibCrowds: surveying our participants

'In the Spotlight' is a crowdsourcing project from the British Library that aims to make digitised historical playbills more discoverable, while also encouraging people to closely engage with this otherwise less accessible collection. Digital Curator Dr Mia Ridge writes...

If you follow our @LibCrowds account on twitter, you might have noticed that we've been working on refreshed versions of our In the Spotlight tasks on Zooniverse. That's part of a small project to enable the use of IIIF manifests on Zooniverse - in everyday language, it means that many, many more digitised items can form the basis of crowdsourcing tasks in the Zooniverse Project Builder, and In the Spotlight is the first project to use this new feature. Along with colleagues in Printed Heritage and BL Labs, I've been looking at our original Pybossa-based LibCrowds site to plan a 'graceful ending' for first phase of the project on LibCrowds.com.

As part of our work documenting and archiving the original LibCrowds site, I'm delighted to share summary results from a 2018 survey of In the Spotlight participants, now published on the British library's Research Repository: https://doi.org/10.23636/w4ee-yc34. Our thanks go to Susan Knight, Customer Insight Coordinator, for her help with the survey.

The survey was designed to help us understand who In the Spotlight participants were, and to help us prioritise work on the project. The 22 question survey was based on earlier surveys run by the Galaxy Zoo and Art UK Tagger projects, to allow comparison with other crowdsourcing projects, and to contribute to our understanding of crowdsourcing in cultural heritage more broadly. It was open to anyone who had contributed to the British Library's In the Spotlight project for historical playbills. The survey was distributed to LibCrowds newsletter subscribers, on the LibCrowds community forum and on social media.

Some headline findings from our survey include:

  • Respondents were most likely to be a woman with a Masters degree, in full-time employment, in London or Southeast UK, who contributes in a break between other tasks or 'whenever they have spare time'.
  • 76% of respondents were motivated by contributing to historical or performance research

Responses to the question 'What was it about this project which caused you to spend more time than intended on it?':

  • Easy to do
  • It's so entertaining
  • Every time an entry is completed you are presented with another item which is interesting and
  • illuminating which provides a continuous temptation regarding what you might discover next
  • simplicity
  • A bit of competitiveness about the top ten contributors but also about contributing something useful
  • I just got carried away with the fun
  • It's so easy to complete
  • Easy to want to do just a few more
  • Addiction
  • Felt I could get through more tasks
  • Just getting engrossed
  • It can be a bit addictive!
  • It's so easy to do that it's very easy to get carried away.
  • interested in the [material]

The summary report contains more rich detail, so go check it out!

 

Crowdsourcing projects from the British Library. 2,969 Volunteers. 265,648 Contributions. 175 Projects
Detail of the front page of libcrowds.com; Crowdsourcing projects from the British Library. 2,969 Volunteers. 265,648 Contributions. 175 Projects