'In search of the sweet spot: infrastructure at the intersection of cultural heritage and data science'

It's not easy to find the abstracts for presentations within panels on the Digital Humanities 2019 (DH2019) site, so I've shared mine here.

In search of the sweet spot: infrastructure at the intersection of cultural heritage and data science

Mia Ridge, British Library

My slides: https://www.slideshare.net/miaridge/in-search-of-the-sweet-spot-infrastructure-at-the-intersection-of-cultural-heritage-and-data-science

This paper explores some of the challenges and paradoxes in the application of data science methods to cultural heritage collections. It is drawn from long experience in the cultural heritage sector, predating but broadly aligned to the 'OpenGLAM' and 'Collections as Data' movements. Experiences that have shaped this thinking include providing open cultural data for computational use; creating APIs for catalogue and interpretive records, running hackathons, and helping cultural organisations think through the preparation of 'collections as data'; and supervising undergraduate and MSc projects for students of computer science.

The opportunities are many. Cultural heritage institutions (aka GLAMS – galleries, libraries, archives and museums) hold diverse historical, scientific and creative works – images, printed and manuscript works, objects, audio or video – that could be turned into some form of digital 'data' for use in data science and digital humanities research. GLAM staff have expert knowledge about the collections and their value to researchers. Data scientists bring a rigour, specialist expertise and skills, and a fresh perspective to the study of cultural heritage collections.

While the quest to publish cultural heritage records and digital surrogates for use in data science is relatively new, the barriers within cultural organisations to creating suitable infrastructure with others are historically numerous. They include different expectations about the pace and urgency of work, different levels of technical expertise, resourcing and infrastructure, and different goals. They may even include different expectations about what 'data' is – metadata drawn from GLAM catalogues is the most readily available and shared data, but not only is this rarely complete, often untidy and inconsistent (being the work of decades or centuries and many hands over that time), it is also a far cry from datasets rich with images or transcribed text that data scientists might expect.

Copyright, data protection and commercial licensing can limit access to digitised materials (though this varies greatly). 'Orphaned works', where the rights holder cannot be traced in order to licence the use of in-copyright works, mean that up to 40% of some collections, particularly sound or video collections, are unavailable for risk-free use.(2012)

While GLAMs have experimented with APIs, downloadable datasets and SPARQL endpoints, they rarely have the resources or institutional will to maintain and refresh these indefinitely. Records may be available through multi-national aggregators such as Europeana, DPLA, or national aggregators, but as aggregation often requires that metadata is mapped to the lowest common denominator, their value for research may be limited.

The area of overlap between 'computationally interesting problems' and 'solutions useful for GLAMs' may be smaller than expected to date, but collaboration between cultural institutions and data scientists on shared projects in the 'sweet spot' – where new data science methods are explored to enhance the discoverability of collections – may provide a way forward. Sector-wide collaborations like the International Image Interoperability Framework (IIIF, https://iiif.io/) provide modern models for lightweight but powerful standards. Pilot projects with students or others can help test the usability of collection data and infrastructure while exploring the applicability of emerging technologies and methods. It is early days for these collaborations, but the future is bright.

Panel overview

An excerpt from the longer panel description by David Beavan and Barbara McGillivray.

This panel highlights the emerging collaborations and opportunities between the fields of Digital Humanities (DH), Data Science (DS) and Artificial Intelligence (AI). It charts the enthusiastic progress of the Alan Turing Institute, the UK national institute for data science and artificial intelligence, as it engages with cultural heritage institutions and academics from arts, humanities and social sciences disciplines. We discuss the exciting work and learnings from various new activities, across a number of high-profile institutions. As these initiatives push the intellectual and computational boundaries, the panel considers both the gains, benefits, and complexities encountered. The panel latterly turns towards the future of such interdisciplinary working, considering how DS & DH collaborations can grow, with a view towards a manifesto. As Data Science grows globally, this panel session will stimulate new discussion and direction, to help ensure the fields grow together and arts & humanities remain a strong focus of DS & AI. Also so DH methods and practices continue to benefit from new developments in DS which will enable future research avenues and questions.

'The Past, Present and Future of Digital Scholarship with Newspaper Collections'

It's not easy to find the abstracts for presentations within panels on the Digital Humanities 2019 (DH2019) site, so I've shared mine here. The panel was designed to bring together range of interdisciplinary newspaper-based digital humanities and/or data science projects, with 'provocations' from two senior scholars who will provide context for current ambitions, and to start conversations among practitioners.

Short Paper: Living with Machines

Paper authors: Mia Ridge, Giovanni Colavizza with Ruth Ahnert, Claire Austin, David Beavan, Kaspar Beelens, Mariona Coll Ardanuy, Adam Farquhar, Emma Griffin, James Hetherington, Jon Lawrence, Katie McDonough, Barbara McGillivray, André Piza, Daniel van Strien, Giorgia Tolfo, Alan Wilson, Daniel Wilson.

My slides: https://www.slideshare.net/miaridge/living-with-machines-at-the-past-present-and-future-of-digital-scholarship-with-newspaper-collections-154700888

Living with Machines is a five-year interdisciplinary research project, whose ambition is to blend data science with historical enquiry to study the human impact of the industrial revolution. Set to be one of the biggest and most ambitious digital humanities research initiatives ever to launch in the UK, Living with Machines is developing a large-scale infrastructure to perform data analyses on a variety of historical sources, and in so doing provide vital insights into the debates and discussions taking place in response to today’s digital industrial revolution.

Seeking to make the most of a self-described 'radical collaboration', the project will iteratively develop research questions as computational linguists, historians, library curators and data scientists work on a shared corpus of digitised newspapers, books and biographical data (census, birth, death, marriage, etc. records). For example, in the process of answering historical research questions, the project could take advantage of access to expertise in computational linguistics to overcome issues with choosing unambiguous and temporally stable keywords for analysis, previously reported by others (Lansdall-Welfare et al., 2017). A key methodological objective of the project is to 'translate' history research questions into data models, in order to inspect and integrate them into historical narratives. In order to enable this process, a digital infrastructure is being collaboratively designed and developed, whose purpose is to marshal and interlink a variety of historical datasets, including newspapers, and allow for historians and data scientists to engage with them.

In this paper we will present our vision for Living with Machines, focusing on how we plan to approach it, and the ways in which digital infrastructure enables this multidisciplinary exchange. We will also showcase preliminary results from the different research 'laboratories', and detail the historical sources we plan to use within the project.

The Past, Present and Future of Digital Scholarship with Newspaper Collections

Mia Ridge (British Library), Giovanni Colavizza (Alan Turing Institute)

Historical newspapers are of interest to many humanities scholars, valued as sources of information and language closely tied to a particular time, social context and place. Following library and commercial microfilming and, more recently, digitisation projects, newspapers have been an accessible and valued source for researchers. The ability to use keyword searches through more data than ever before via digitised newspapers has transformed the work of researchers.[1]

Digitised historic newspapers are also of interest to many researchers who seek large bodies of relatively easily computationally-transcribed text on which they can try new methods and tools. Intensive digitisation over the past two decades has seen smaller-scale or repository-focused projects flourish in the Anglophone and European world (Holley, 2009; King, 2005; Neudecker et al., 2014). However, just as earlier scholarship was potentially over-reliant on The Times of London and other metropolitan dailies, this has been replicated and reinforced by digitisation projects (for a Canadian example, see Milligan 2013).

In the last years, several large consortia projects proposing to apply data science and computational methods to historical newspapers at scale have emerged, including NewsEye, impresso, Oceanic Exchanges and Living with Machines. This panel has been convened by some consortia members to cast a critical view on past and ongoing digital scholarship with newspapers collections, and to inform its future.

Digitisation can involve both complexities and simplifications. Knowledge about the imperfections of digitisation, cataloguing, corpus construction, text transcription and mining is rarely shared outside cultural institutions or projects. How can these imperfections and absences be made visible to users of digital repositories? Furthermore, how does the over-representation of some aspects of society through the successive winnowing and remediation of potential sources – from creation to collection, microfilming, preservation, licensing and digitisation – affect scholarship based on digitised newspapers. How can computational methods address some of these issues?

The panel proposes the following format: short papers will be delivered by existing projects working on large collections of historical newspapers, presenting their vision and results to date. Each project is at different stages of development and will discuss their choice to work with newspapers, and reflect on what have they learnt to date on practical, methodological and user-focused aspects of this digital humanities work. The panel is additionally an opportunity to consider important questions of interoperability and legacy beyond the life of the project. Two further papers will follow, given by scholars with significant experience using these collections for research, in order to provide the panel with critical reflections. The floor will then open for debate and discussion.

This panel is a unique opportunity to bring senior scholars with a long perspective on the uses of newspapers in scholarship together with projects at formative stages. More broadly, convening this panel is an opportunity for the DH2019 community to ask their own questions of newspaper-based projects, and for researchers to map methodological similarities between projects. Our hope is that this panel will foster a community of practice around the topic and encourage discussions of the methodological and pedagogical implications of digital scholarship with newspapers.

[1] For an overview of the impact of keyword search on historical research see (Putnam, 2016) (Bingham, 2010).

Notes from 'AI, Society & the Media: How can we Flourish in the Age of AI'

Before we start: in the spirit of the mid-2000s, I thought I'd have a go at blogging about events again. I've realised I miss the way that blogging and reading other people's posts from events made me feel part of a distributed community of fellow travellers. Journal articles don't have the same effect (they're too long and jargony for leisure readers, assuming they're accessible outside universities at all), and tweets are great for connecting with people, but they're very ephemeral. Here goes…

BBC Broadcasting House

On September 3 I was at BBC Broadcasting House for 'AI, Society & the Media: How can we Flourish in the Age of AI?' by BBC, LCFI and The Alan Turing Institute. Artificial intelligence is a hot topic so it was a sell-out event. My notes are very partial (in both senses of the word), and please do let me know if there are errors. The event hashtag will provide more coverage: https://twitter.com/hashtag/howcanweflourish.

The first session was 'AI – What you need to know!'. Matthew Postgate began by providing context for the BBC's interest in AI. 'We need a plurality of business models for AI – not just ad-funded' – yes! The need for different models for AI (and related subjects like machine learning) was a theme that recurred throughout the day (and at other events I was at this week).

Adrian Weller spoke on the limitations of AI. It's data hungry, compute intensive, poor at representing uncertainty, easily fooled by adversarial examples (and more that I missed). We need sensible measures of trustworthiness including robustness, fairness, protection of privacy, transparency.

Been Kim shared Google's AI principles: https://ai.google/principles She's focused on interpretability – goals are to ensure that our values are aligned and our knowledge is reflected. She emphasised the need to understand your data (another theme across the day and other events this week). You can an inherently interpretable machine model (so it can explain its reasoning) or can build an interpreter, enabling conversations between humans and machines. You can then uncover bias using the interpreter, asking what weight it gave to different aspects in making decisions.

Jonnie Penn (who won me with an early shout out to the work of Jon Agar) asked, from where does AI draw its authority? AI is feeding a monopoly of Google-Amazon-Facebook who control majority of internet traffic and advertising spend. Power lies in choosing what to optimise for, and choosing what not to do (a tragically poor paraphrase of his example of advertising to children, but you get the idea). We need 'bureaucratic biodiversity' – need lots of models of diverse systems to avoid calcification.

Kate Coughlan – only 10% of people feel they can influence AI. They looked at media narratives re AI on axes of time (ease vs obsolescence), power (domination vs uprising), desire (gratification vs alienation), life (immortality vs inhumanity). Their survey found that each aspect was equally disempowering. Passivity drives negative outcomes re feelings about change, tech – but if people have agency, then it's different. We need to empower citizens to have active role in shaping AI.

The next session was 'Fake News, Real Problems: How AI both builds and destroys trust in news'. Ryan Fox spoke on 'manufactured consensus' – we're hardwired to agree with our community so you can manipulate opinion by making it look like everyone else thinks a certain way. Manipulating consensus is currently legal, though against social network T&S. 'Viral false narratives can jeopardise brand trust and integrity in an instant'. Manufactured outrage campaigns etc. They're working on detecting inorganic behaviour through the noise – it's rapid, repetitive, sticky, emotional (missed some).

One of the panel questions was, would AI replace journalists? No, it's more like having lots of interns – you wouldn't have them write articles. AI is good for tasks you can explain to a smart 16 year old in the office for a day. The problematic ad-based model came up again – who is the arbiter of truth (e.g. fake news on Facebook). Who's paying for those services and what power does it give them?

This panel made me think about discussions about machine learning and AI at work. There are so many technical, contextual and ethical challenges for collecting institutions in AI, from capturing the output of an interactive voice experience with Alexa, to understanding and recording the difference between Russia Today as a broadcast news channel and as a manipulator of YouTube rankings.

Next was a panel on 'AI as a Creative Enabler'. Cassian Harrison spoke about 'Made By Machine', an experiment with AI and archive programming. They used scene detection, subtitle analysis, visual 'energy', machine learning on the BBC's Redux archive of programmes. Programmes were ranked by how BBC4 they were; split into sections then edited down to create mini BBC4 programmes.

Kanta Dihal and Stephen Cave asked why AI fascinates us in a thoughtful presentation. It's between dead and alive, uncanny (and lots more but clearly my post-lunch notetaking isn't the best).

Anna Ridler and Amy Cutler have created an AI-scripted nature documentary (trained on and re-purposing a range of tropes and footage from romance novels and nature documentaries) and gave a brilliant presentation about AI as a medium and as a process. Anna calls herself a dataset artist, rather than a machine learning artist. You need to get to know the dataset, look out for biases and mistakes, understand the humanness of decisions about what was included or excluded. Machines enact distorted versions of language.

Text from slide is transcribed above
Diane Coyle on 'Lessons for the era of AI'

I don't have notes from 'Next Gen AI: How can the next generation flourish in the age of AI?' but it was great to hear about hackathons where teenagers could try applying AI. The final session was 'The Conditions for Flourishing: How to increase citizen agency and social value'. Hannah Fry – once something is dressed up as an algorithm it gains some authority that's hard to question. Diane Coyle talked about 'general purpose technologies', which transform one industry then others. Printing, steam, electricity, internal combustion engine, digital computing, AI. Her 'lessons for the era of AI' were: all technology is social; all technologies are disruptive and have unpredictable consequences; all successful technologies enhance human freedoms', and accordingly she suggested we 'think in systems; plan for change; be optimistic'.

Konstantinos Karachalios called for a show of hands re who feels they have control over their data and what's done with it? Very few hands were raised. 'If we don't act now we'll lose our agency'.

I'm going to give the final word to Terah Lyons as the key takeaway from the day: 'technology is not destiny'.

I didn't hear a solution to the problems of 'fake news' that doesn't require work from all of us. If we don't want technology to be destiny, we all need pay attention to the applications of AI in our lives, and be prepared to demand better governance and accountability from private and government agents.

(A bonus 'question I didn't ask' for those who've read this far: how do BBC aims for ethical AI relate to the introduction compulsory registration to access tv and radio? If I turn on the radio in my kitchen, my listening habits aren't tracked; if I listen via the app they're linked to my personal ID).

Updates from Digital Scholarship at the British Library

I've been posting on the work blog far more frequently than I have here. Launching and running In the Spotlight, crowdsourcing the transcription of the British Library's historic playbills collection, was a focus in 2017-18. Some blog posts:

And a press release and newsletters:

Other updates from work, including a new project, information about the Digital Scholarship Reading Group I started, student projects, and an open data project I shepherded:

Cross-post: Seeking researchers to work on an ambitious data science and digital humanities project

I rarely post here at the moment, in part because I post on the work blog. Here's a cross-post to help spread the word about some exciting opportunities currently available: Seeking researchers to work on an ambitious data science and digital humanities project at the British Library and Alan Turing Institute (London)

'If you follow @BL_DigiSchol or #DigitalHumanities hashtags on twitter, you might have seen a burst of data science, history and digital humanities jobs being advertised. In this post, Dr Mia Ridge of the Library's Digital Scholarship team provides some background to contextualise the jobs advertised with the 'Living with Machines' project.

We are seeking to appoint several new roles who will collaborate on an exciting new project developed by the British Library and The Alan Turing Institute, the national centre for data science and artificial intelligence.

Jobs currently advertised:

The British Library jobs are now advertised, closing September 21:

You may have noticed that the British Library is also currently advertising for a Curator, Newspaper Data (closes Sept 9). This isn’t related to Living with Machines, but with an approach of applying data-driven journalism and visualisation techniques to historical collections, it should have some lovely synergies and opportunities to share work in progress with the project team. There's also a Research Software Engineer advertised that will work closely with many of the same British Library teams.

If you're applying for these posts, you may want to check out the Library's visions and values on the refreshed 'Careers' website.'

My opening remarks for MCG's Museums+Tech 2017

My notes introducing the theme of the Museums Computer Group's 2017 conference and a call to action for people working in cultural heritage technology below.

A divided world

2016 was the year that deep fractures came to the surface, but they’d been building for some time. We might live in the same country as each other, but we can experience it very differently. What we know about the state of the world is affected by where we live, our education, and by how (if?) we get our news.

Life in 2017

Cartoon of a dog surrounded by fire drinking coffee

    'This is fine' (KC Green)

We can't pretend that it'll all go away and that society will heal itself. Divisions over Brexit, the role of propaganda in elections, climate change, the role of education, what we value as a society – they're all awkward to address, but if we don't it's hard to see how we can move forward. And since we're here to talk about museums – what role do museums have in divided societies? How much do they need to reflect voices they mightn't agree with? Do we need to make ourselves a bit uncomfortable in order to make spaces for sharing experiences and creating empathy? Can (digital) experiences, collections and exhibitions in cultural heritage help create a shared understanding of the world?

'arts and cultural engagement [helps] shape reflective individuals, facilitating greater understanding of themselves and their lives, increasing empathy with respect to others, and an appreciation of the diversity of human experience and cultures.' From Understanding the value of arts & culture: The AHRC Cultural Value Project by Geoffrey Crossick & Patrycja Kaszynska

I've been struck lately by the observation that empathy can bridge divides, and give people the power to understand others. The arts and culture provide opportunities to 'understand and share in another person's feelings and experiences' and connect the past to the present. How can museums – in all their different forms – contribute to a more empathic (and maybe eventually less divided) society?

'The greatest benefit we owe to the artist, whether painter, poet, or novelist, is the extension of our sympathies. … Art is the nearest thing to life; it is a mode of amplifying experience and extending our contact with our fellow-men beyond the bounds of our personal lot.' George Eliot, as quoted in Peter Bazalgette's The Empathy Instinct

Digital experiences aren't shared in the same way as physical ones, and ‘social’ media isn't the same as being in the same space as someone experiencing the same thing, but they have other advantages – I hope we'll learn about some today.

We need to tell better stories about museums and computers

Woman with buckets of computer cables
Engineer Karen Leadlay in Analog Computer Lab

Shifting from the public to staff in museums… Museums have been using technology to serve audiences and manage collections for decades. But still it feels like museums are criticised for simultaneously having too much and too little technology. Shiny apps make the news, but they're built on decades of digitisation and care from heritage organisations. There's a lot museums could do better, and digital expertise is not evenly distributed or recognised, but there's a lot that's done well, too. My challenge to you is to find and share better stories about cultural heritage technologies connecting collections, people and knowledge. If we don't tell those stories, they'll be told about us. Too many articles and puff pieces ignore the thoughtful, quotidian and/or experimental work of experts across the digital cultural heritage sector.

[Later in the day I mentioned that the conference had an excellent response to the call for papers – we learnt about more interesting projects than we had room to fit in, so perhaps we should encourage more people to post case studies to the MCG's discussion list and website.]

The Museums+Tech 2017 programme

  • Keynote: ‘What makes a Museum?
  • Museums in a post-truth world of fake news
  • Challenging Expectations
  • Dealing with distance; bringing the museum to the people
  • How can museums use sound and chatbots?
  • Looking (back to look) forward

Speaking of better stories – I'm looking forward to hearing from all our speakers today – they're covering an incredible range of topics, approaches and technologies, so hopefully each of you will leave full of ideas. Join us for drinks afterwards to keep the conversation going. And to set the tone for the day, it's a great time to hear Hannah Fox on the topic of 'what makes a museum'

Speaking of the conference – a lot of people helped out in different ways, so thanks to them all!

Save

Save

Save

From piles of material to patchwork: How do we embed the production of usable collections data into library work?

How do we embed the production of usable collections data into library work?These notes were prepared for a panel discussion at the 'Always Already Computational: Collections as Data' (#AACdata) workshop, held in Santa Barbara in March 2017. While my latest thinking on the gap between the scale of collections and the quality of data about them is informed by my role in the Digital Scholarship team at the British Library, I've also drawn on work with catalogues and open cultural data at Melbourne Museum, the Museum of London, the Science Museum and various fellowships. My thanks to the organisers and the Institute of Museum and Library Services for the opportunity to attend. My position paper was called 'From libraries as patchwork to datasets as assemblages?' but in hindsight, piles and patchwork of material seemed a better analogy.

The invitation to this panel asked us to share our experience and perspective on various themes. I'm focusing on the challenges in making collections available as data, based on years of working towards open cultural data from within various museums and libraries. I've condensed my thoughts about the challenges down into the question on the slide: How do we embed the production of usable collections data into library work?

It has to be usable, because if it's not then why are we doing it? It has to be embedded because data in one-off projects gets isolated and stale. 'Production' is there because infrastructure and workflow is unsexy but necessary for access to the material that makes digital scholarship possible.

One of the biggest issues the British Library (BL) faces is scale. The BL's collections are vast – maybe 200 million items – and extremely varied. My experience shows that publishing datasets (or sharing them with aggregators) exposes the shortcomings of past cataloguing practices, making the size of the backlog all too apparent.

Good collections data (or metadata, depending on how you look at it) is necessary to avoid the overwhelmed, jumble sale feeling of using a huge aggregator like Europeana, Trove, or the DPLA, where you feel there's treasure within reach, if only you could find it. Publishing collections online often increases the number of enquiries about them – how can institution deal with enquiries at scale when they already have a cataloguing backlog? Computational methods like entity identification and extraction could complement the 'gold standard' cataloguing already in progress. If they're made widely available, these other methods might help bridge the resourcing gaps that mean it's easier to find items from richer institutions and countries than from poorer ones.

Photo of piles of materialYou probably already all know this, but it's worth remembering: our collections aren't even (yet) a patchwork of materials. The collections we hold, and the subset we can digitise and make available for re-use are only a tiny proportion of what once existed. Each piece was once part of something bigger, and what we have now has been shaped by cumulative practical and intellectual decisions made over decades or centuries. Digitisation projects range from tiny specialist databases to huge commercial genealogy deals, while some areas of the collections don't yet have digital catalogue records. Some items can't be digitised because they're too big, small or fragile for scanning or photography; others can't be shared because of copyright, data protection or cultural sensitivities. We need to be careful in how we label datasets so that the absences are evident.

(Here, 'data' may include various types of metadata, automatically generated OCR or handwritten text recognition transcripts, digital images, audio or video files, crowdsourced enhancements or any combination or these and more)

Image credit: https://www.flickr.com/photos/teen_s/6251107713/

In addition to the incompleteness or fuzziness of catalogue data, when collections appear as data, it's often as great big lumps of things. It's hard for normal scholars to process (or just unzip) 4gb of data.

Currently, datasets are often created outside normal processes, and over time they become 'stale' as they're not updated when source collections records change. And when they manage to unzip them, the records rely on internal references – name authorities for people, places, etc – that can only be seen as strings rather than things until extra work is undertaken.

The BL's metadata team have experimented with 'researcher format' CSV exports around specific themes (eg an exhibition), and CSV is undoubtedly the most accessible format – but what we really need is the ability for people to create their own queries across catalogues, and create their own datasets from the results. (And by queries I don't mean SPARQL but rather faceted browsing or structured search forms).

Image credit: screenshot from http://data.bl.uk/

Collections are huge (and resources relatively small) so we need to supplement manual cataloguing with other methods. Sometimes the work of crafting links from catalogues to external authorities and identifiers will be a machine job, with pieces sewn together at industrial speed via entity recognition tools that can pull categories out or text and images. Sometimes it's operated by a technologist who runs records through OpenRefine to find links to name authorities or Wikidata records. Sometimes it's a labour of scholarly love, with links painstakingly researched, hand-tacked together to make sure they fit before they're finally recorded in a bespoke database.

This linking work often happens outside the institution, so how can we ingest and re-use it appropriately? And if we're to take advantage of computational methods and external enhancements, then we need ways to signal which categories were applied by catalogues, which by software, by external groups, etc.

The workflow and interface adjustments required would be significant, but even more challenging would be the internal conversations and changes required before a consensus on the best way to combine the work of cataloguers and computers could emerge.

The trick is to move from a collection of pieces to pieces of a collection. Every collection item was created in and about places, and produced by and about people. They have creative, cultural, scientific and intellectual properties. There's a web of connections from each item that should be represented when they appear in datasets. These connections help make datasets more usable, turning strings of text into references to things and concepts to aid discoverability and the application of computational methods by scholars. This enables structured search across datasets – potentially linking an oral history interview with a scientist in the BL sound archive, their scientific publications in journals, annotated transcriptions of their field notebooks from a crowdsourcing project, and published biography in the legal deposit library.

A lot of this work has been done as authority files like AAT, ULAN etc are applied in cataloguing, so our attention should turn to turning local references into URIs and making the most of that investment.

Applying identifiers is hard – it takes expert care to disambiguate personal names, places, concepts, even with all the hinting that context-aware systems might be able to provide as machine learning etc techniques get better. Catalogues can't easily record possible attributions, and there's understandable reluctance to publish an imperfect record, so progress on the backlog is slow. If we're not to be held back by the need for records to be perfectly complete before they're published, then we need to design systems capable of capturing the ambiguity, fuzziness and inherent messiness of historical collections and allowing qualified descriptors for possible links to people, places etc. Then we need to explain the difference to users, so that they don't overly rely on our descriptions, making assumptions about the presence or absence of information when it's not appropriate.

Image credit: http://europeana.eu/portal/record/2021648/0180_N_31601.html

Photo of pipes over a buildingA lot of what we need relies on more responsive infrastructure for workflows and cataloguing systems. For example, the BL's systems are designed around the 'deliverable unit' – the printed or bound volume, the archive box – because for centuries the reading room was where you accessed items. We now need infrastructure that makes items addressable at the manuscript, page and image level in order to make the most of the annotations and links created to shared identifiers.

(I'd love to see absorbent workflows, soaking up any related data or digital surrogates that pass through an organisation, no matter which system they reside in or originate from. We aren't yet making the most of OCRd text, let alone enhanced data from other processes, to aid discoverability or produce datasets from collections.)

Image credit: https://www.flickr.com/photos/snorski/34543357
My final thought – we can start small and iterate, which is just as well, because we need to work on understanding what users of collections data need and how they want to use them. We're making a start and there's a lot of thoughtful work behind the scenes, but maybe a bit more investment is needed from research libraries to become as comfortable with data users as they are with the readers who pass through their physical doors.

Trying computational data generation and entity extraction for images and text

I've developed this exercise on computational data generation and entity extraction for various information/data visualisation workshops I've been teaching lately. These exercises help demonstrate the biases embedded in machine learning and 'AI' tools. As these methods have become more accessible, my dataviz workshops have included more discussion of computational methods for generating data to be visualised. There are two versions of the exercise – the first works with images, the second with text.

In teaching I've found that services that describe images were more accessible and generated richer discussion in class than text-based sites, but it's handy to have the option for people who work with text. If you try something like this in your classes I'd love to hear from you.

It's also a chance to talk about the uses of these technologies in categorising and labelling our posts on social media. We can tell people that their social media posts are analysed for personality traits and mentions of brands, but seeing it in action is much more powerful.

Image exercise: trying computational data generation and entity extraction

Time: c. 5 – 10 minutes plus discussion.

Goal: explore methods for extracting information from text or an image and reflect on what the results tell you about the algorithms

1. Find a sample image

Find an image (e.g. from a news site or digitised text) you can download and drag into the window. It may be most convenient to save a copy to your desktop. Many sites let you load images from a URL, so right- or control-clicking to copy an image location for pasting into the site can be useful.

2. Work in your browser

It's probably easiest to open each of these links in a new browser window. It's best to use Firefox or Chrome, if you can. Safari and Internet Explorer may behave slightly differently on some sites. You should not need to register to use these sites – please read the tips below or ask for help if you get stuck.

3. Review the outputs

Make notes, or discuss with your neighbour. Be prepared to report back to the group.

  • What attributes does each tool report on?
  • Which attributes, if any, were unique to a service?
  • Based on this, what do companies like Clarifai, Google, IBM and Microsoft seem to think is important to them (or to their users)? (e.g. what does 'safe for work' really mean?)
  • Who are their users – the public or platform administrators?
  • How many of possible entities (concepts, people, places, events, references to time or dates, etc) did it pick up?
  • Is any of the information presented useful?
  • Did it label anything incorrectly?
  • What options for exporting or saving the results did the demo offer? What about the underlying service or software?
  • For tools with configuration options – what could you configure? What difference did changing classifiers or other parameters  make?
  • If you tried it with a few images, did it do better with some than others? Why might that be?

Text exercise: trying computational data generation and entity extraction

Time: c. 5 minutes plus discussion
Goal: explore the impact of source data and algorithms on input text

1.     Grab some text

You will need some text for this exercise. The more 'entities' – people, places, dates, concepts – discussed, the better. If you have some text you're working on handy, you can use that. If you're stuck for inspiration, pick a front page story from an online news site. Keep the page open so you can copy a section of text to paste into the websites.

2.     Compare text entity labelling websites

  • Open four or more browser windows or tabs. Open the links below in separate tabs or windows so you can easily compare the results.
  • Go to DBpedia Spotlight https://dbpedia-spotlight.github.io/demo/. Paste your copied text into the box, or keep the sample text in the box. Hit 'Annotate'.
  • Go to Ontotext http://tag.ontotext.com/. You may need to click through the opening screen. Paste your copied text into the box. Hit 'annotate'.
  • Finally, go to Stanford Named Entity Tagger http://nlp.stanford.edu:8080/ner/. Paste your text into the box. Hit 'Submit query'.

3.     Review the outputs

  • How many possible entities (concepts, people, places, events, references to time or dates, etc) did each tool pick up? Is any of the other information presented useful?
  • Did it label anything incorrectly?
  • What if you change classifiers or other parameters?
  • Does it do better with different source material?
  • What differences did you find between the two tools? What do you think caused those differences?
  • How much can you find out about the tools and the algorithms they use to create labels?
  • Where does the data underlying the process come from?

Spoiler alert!

screenshot
Clarifai's image recognition tool with a historical image

Network visualisations and the 'so what?' problem

This week I was in Luxembourg for a workshop on Network Visualisation in the Cultural Heritage Sector, organised by Marten Düring and held on the Belval campus of the University of Luxembourg.

In my presentation, I responded to some of the questions posed in the workshop outline:

In this workshop we want to explore how network visualisations and infrastructures will change the research and outreach activities of cultural heritage professionals and historians. Among the questions we seek to discuss during the workshop are for example: How do users benefit from graphs and their visualisation? Which skills do we expect from our users? What can we teach them? Are SNA [social network analysis] theories and methods relevant for public-facing applications? How do graph-based applications shape a user’s perception of the documents/objects which constitute the data? How can applications benefit from user engagement? How can applications expand and tap into other resources?

A rough version of my talk notes is below. The original slides are also online.

Network visualisations and the 'so what?' problem

Caveat

While I may show examples of individual network visualisations, this talk isn't a critique of them in particular. There's lots of good practice around, and these lessons probably aren't needed for people in the room.

Fundamentally, I think network visualisations can be useful for research, but to make them more effective tools for outreach, some challenges should be addressed.

Context

I'm a Digital Curator at the British Library, mostly working with pre-1900 collections of manuscripts, printed material, maps, etc. Part of my job is to help people get access to our digital collections. Visualisations are a great way to firstly help people get a sense of what's available, and then to understand the collections in more depth.

I've been teaching versions of an 'information visualisation 101' course at the BL and digital humanities workshops since 2013. Much of what I'm saying now is based on comments and feedback I get when presenting network visualisations to academics, cultural heritage staff (who should be a key audience for social network analyses).

Provocation: digital humanists love network visualisations, but ordinary people say, 'so what'?

Fig1And this is a problem. We're not conveying what we're hoping to convey.

Network visualisation, via Table of data, via http://fredbenenson.com/
Network visualisation http://fredbenenson.com

When teaching datavis, I give people time to explore examples like this, then ask questions like 'Can you tell what is being measured or described? What do the relationships mean?'. After talking about the pros and cons of network visualisations, discussion often reaches a 'yes, but so what?' moment.

Here are some examples of problems ordinary people have with network visualisations…

Location matters

Spatial layout based on the pragmatic aspects of fitting something on the screen using physics, rules of attraction and repulsion doesn't match what people expect to see. It's really hard for some to let go of the idea that spatial layout has meaning. The idea that location on a page has meaning of some kind is very deeply linked to their sense of what a visualisation is.

Animated physics is … pointless?

People sometimes like the sproinginess when a network visualisation resettles after a node has been dragged, but waiting for the animation to finish can also be slow and irritating. Does it convey meaning? If not, why is it there?

Size, weight, colour = meaning?

The relationship between size, colour, weight isn't always intuitive – people assume meaning where there might be none.

In general, network visualisations are more abstract than people expect a visualisation to be.

'What does this tell me that I couldn't learn as quickly from a sentence, list or table?'

Table of data, via http://fredbenenson.com/
Table of data, via http://fredbenenson.com/

Scroll down the page that contains the network graph above and you get other visualisations. Sometimes they're much more positively received, particularly people feel they learn more from them than from the network visualisation.

Onto other issues with 'network visualisations as communication'…

Which algorithmic choices are significant?

screenshot of network graphs
Mike Bostock's force-directed and curved line versions of character co-occurrence in Les Misérables

It's hard for novices to know which algorithmic and data-cleaning choices are significant, and which have a more superficial impact.

Untethered images

Images travel extremely well on social media. When they do so, they often leave information behind and end up floating in space. Who created this, and why? What world view does it represent? What source material underlies it, how was it manipulated to produce the image? Can I trust it?

'Can't see the wood for the trees'

viral texts

When I showed this to a class recently, one participant was frustrated that they couldn't 'see the wood for the trees'. The visualisations gives a general impression of density, but it's not easy to dive deeper into detail.

Stories vs hairballs

But when I started to explain what was being represented – the ways in which stories were copied from one newspaper to another – they were fascinated. They might have found their way there if they'd read the text but again, the visualisation is so abstract that it didn't hint at what lay underneath. (Also I have only very, very rarely seen someone stop to read the text before playing with a visualisation.)

No sense of change over time

This flattening of time into one simultaneous moment is more vital for historical networks than for literary ones, but even so, you might want to compare relationships between sections of a literary work.

No sense of texture, detail of sources

All network visualisations look similar, whether they're about historical texts or cans of baked beans. Dots and lines mask texture, and don't always hint at the depth of information they represent.

Jargon

Node. Edge. Graph. Directed, undirected. Betweenness. Closeness. Eccentricity.

There's a lot to take on to really understand what's being expressed in a network graph.

There is some hope…

Onto the positive bit!

Interactivity is engaging

People find the interactive movement, the ability to zoom and highlight links engaging, even if they have no idea what's being expressed. In class, people started to come up with questions about the data as I told them more about what was represented. That moment of curiosity is an opportunity if they can dive in and start to explore what's going on, what do the relationships mean?

…but different users have different interaction needs

For some, there's that frustration expressed earlier they 'can't get to see a particular tree' in the dense woods of a network visualisation. People often want to get to the detail of an instance of a relationship – the lines of text, images of the original document – from a graph.

This mightn't be how network visualisations are used in research, but it's something to consider for public-facing visualisations. How can we connect abstract lines or dots to detail, or provide more information about what the relationship means, show the quantification expressed as people highlight or filter parts of a graph? A  harder, but more interesting task is hinting at the texture or detail of those relationships.

Proceed, with caution

One of the workshop questions was 'Are social network analysis theories and methods relevant for public-facing applications?' – and maybe the answer is a qualified yes. As a working tool, they're great for generating hypotheses, but they need a lot more care before exposing them to the public.

[As an aside, I’d always taken the difference between visualisations as working tools for exploring data – part of the process of investigating a research question – and visualisation as an output – a product of the process, designed for explanation rather than exploration – as fundamental, but maybe we need to make that distinction more explicit.]

But first – who are your 'users'?

During this workshop, at different points we may be talking about different 'users' – it's useful to scope who we mean at any given point. In this presentation, I was talking about end users who encounter visualisations, not scholars who may be organising and visualising networks for analysis.

Sometimes a network visualisation isn't the answer … even if it was part of the question.

As an outcome of an exploratory process, network visualisations are not necessarily the best way to present the final product. Be disciplined – make yourself justify the choice to use network visualisations.

No more untethered images

Include an extended caption – data source, tools and algorithms used. Provide a link to find out more – why this data, this form? What was interesting but not easily visualised? Let people download the dataset to explore themselves?

Present visualisations as the tip of the data iceberg

Visualisations are the tip of the iceberg
Visualisations are the tip of the iceberg

Lots of interesting data doesn't make it into a visualisation. Talking about what isn't included and why it was left out is important context.

Talk about data that couldn't exist

Beyond the (fuzzy, incomplete, messy) data that's left out because it's hard to visualise, data that never existed in the first place is also important:

'because we're only looking on one axis (letters), we get an inflated sense of the importance of spatial distance in early modern intellectual networks. Best friends never wrote to each other; they lived in the same city and drank in the same pubs; they could just meet on a sunny afternoon if they had anything important to say. Distant letters were important, but our networks obscure the equally important local scholarly communities.'
Scott Weingart, 'Networks Demystified 8: When Networks are Inappropriate'

Help users learn the skills and knowledge they need to interpret network visualisations in context.

How? Good question! This is the point at which I hand over to you…

SXSW, project anniversaries and more – news on heritage crowdsourcing

Photo of programme
Our panel listing at SXSW

I've just spent two weeks in Texas, enjoying the wonderful hospitality and probing questions after giving various talks at universities in Houston and Austin before heading to SXSW. I was there for a panel on 'Build the Crowdsourcing Community of Your Dreams' (link to our slides and collected resources) with Ben Brumfield, Siobhan Leachman, and Meghan Ferriter. Siobhan, a 'super-volunteer' in more ways than one, posted her talk notes on 'How cultural institutions encouraged me to participate in crowdsourcing & the factors I consider before donating my time'.

In other news, we (me, Ben, Meghan and Christy Henshaw from the Wellcome Library) have had a workshop accepted for the Digital Humanities 2016 conference, to be held in Kraków in July. We're looking for people with different kinds of expertise for our DH2016 Expert Workshop: Beyond The Basics: What Next For Crowdsourcing?.  You can apply via this form.

One of the questions at our SXSW panel was about crowdsourcing in teaching, which reminded me of this recent post on 'The War Department in the Classroom' in which Zayna Bizri 'describes her approach to using the Papers of the War Department in the classroom and offers suggestions for those who wish to do the same'. In related news, the PWD project is now five years old! There's also this post on Primary School Zooniverse Volunteers.

The Science Gossip project is one year old, and they're asking their contributors to decide which periodicals they'll work on next and to start new discussions about the documents and images they find interesting.

The History Harvest project have released their Handbook (PDF).

The Danish Nationalmuseet is having a 'Crowdsource4dk' crowdsourcing event on April 9. You can also transcribe Churchill's WWII daily appointments, 1939 – 1945 or take part in Old Weather: Whaling (and there's a great Hyperallergic post with lots of images about the whaling log books).

I've seen a few interesting studentships and jobs posted lately, hinting at research and projects to come. There's a funded PhD in HCI and online civic engagement and a (now closed) studentship on Co-creating Citizen Science for Innovation.

And in old news, this 1996 post on FamilySearch's collaborative indexing is a good reminder that very little is entirely new in crowdsourcing.