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.

Helping us fly? Machine learning and crowdsourcing

Image of a man in a flying contrapation powered by birds
Moon Machine by Bernard Brussel-Smith via Serendip-o-matic

Over the past few years we've seen an increasing number of projects that take the phrase 'human-computer interaction' literally (perhaps turning 'HCI' into human-computer integration), organising tasks done by people and by computers into a unified system. One of the most obvious benefits of crowdsourcing on digital platforms has been the ability to coordinate the distribution and validation of tasks. Increasingly, data manually classified through crowdsourcing is being fed into computers to improve machine learning so that computers can learn to recognise images or words almost as well as we do. I've outlined a few projects putting this approach to work below.

This creates new challenges for the future: if fun, easy tasks like image tagging and text transcription can be done by computers, what are the implications for cultural heritage and digital humanities crowdsourcing projects that used simple tasks as the first step in public engagement? After all, Fast Company reported that 'at least one Zooniverse project, Galaxy Zoo Supernova, has already automated itself out of existence'. What impact will this have on citizen science and history communities? How might machine learning free us to fly further, taking on more interesting tasks with cultural heritage collections?

The Public Catalogue Foundation has taken tags created through Your Paintings Tagger and achieved impressive results in the art of computer image recognition: 'Using the 3.5 million or so tags provided by taggers, the research team at Oxford 'educated' image-recognition software to recognise the top tagged terms'. All paintings tagged with a particular subject (e.g. 'horse') were fed into feature extraction processes to build an 'object model' of a horse (a set of characteristics that would indicate that a horse is depicted) then tested to see the system could correctly tag horses.

The BBC World Service archive used an 'open-source speech recognition toolkit to listen to every programme and convert it to text' and keywords then asked people to check the correctness of the data created (Algorithms and Crowd-Sourcing for Digital Archives, see also What we learnt by crowdsourcing the World Service archive).

The CUbRIK project combines 'machine, human and social computation for multimedia search' in their technical demonstrator, HistoGraph. The SOCIAM: The Theory and Practice of Social Machines project is looking at 'a new kind of emergent, collective problem solving', including 'citizen science social machines'.

And of course the Zooniverse is working on this, most recently with Galaxy Zoo. A paper summarised on their Milky Way project blog, outlines the powerful synergy between citizens scientists, professional scientists, and machine learning: 'citizens can identify patterns that machines cannot detect without training, machine learning algorithms can use citizen science projects as input training sets, creating amazing new opportunities to speed-up the pace of discovery', addressing the weakness of each approach if deployed alone.

Further reading: an early discussion of human input into machine learning is in Quinn and Bederson's 2011 Human Computation: A Survey and Taxonomy of a Growing Field. You can get a sense of the state of the field from various conference papers, including ICML ’13 Workshop: Machine Learning Meets Crowdsourcing and ICML ’14 Workshop: Crowdsourcing and Human Computing. There's also a mega-list of academic crowdsourcing conferences and workshops, though it doesn't include much on the tiny corner of the world that is crowdsourcing in cultural heritage.

Last update: March 2015. This post collects my thoughts on machine learning and human-computer integration as I finish my thesis. Do you know of examples I've missed, or implications we should consider?

Slow and still dirty Digital Humanities Australasia notes: day 3

These are my very rough notes from day 3 of the inaugural Australasian Association for Digital Humanities conference (see also Quick and dirty Digital Humanities Australasia notes: day 1 and Quick and dirty Digital Humanities Australasia notes: day 2) held in Canberra's Australian National University at the end of March.

We were welcomed to Day 3 by the ANU's Professor Marnie Hughes-Warrington (who expressed her gratitude for the methodological and social impact of digital humanities work) and Dr Katherine Bode.  The keynote was Dr Julia Flanders on 'Rethinking Collections', AKA 'in praise of collections'… [See also Axel Brun's live blog.]

She started by asking what we mean by a 'collection'? What's the utility of the term? What's the cultural significance of collections? The term speaks of agency, motive, and implies the existence of a collector who creates order through selectivity. Sites like eBay, Flickr, Pinterest are responding to weirdly deep-seated desire to reassert the ways in which things belong together. The term 'collection' implies that a certain kind of completeness may be achieved. Each item is important in itself and also in relation to other items in the collection.

There's a suite of expected activities and interactions in the genre of digital collections, projects, etc. They're deliberate aggregations of materials that bear, demand individual scrutiny. Attention is given to the value of scale (and distant reading) which reinforces the aggregate approach…

She discussed the value of deliberate scope, deliberate shaping of collections, not craving 'everythingness'. There might also be algorithmically gathered collections…

She discussed collections she has to do with – TAPAS, DHQ, Women Writers Online – all using flavours of TEI, the same publishing logic, component stack, providing the same functionality in the service of the same kinds of activities, though they work with different materials for different purposes.

What constitutes a collection? How are curated collections different to user-generated content or just-in-time collections? Back 'then', collections were things you wanted in your house or wanted to see in the same visit. What does the 'now' of collections look like? Decentralisation in collections 'now'… technical requirements are part of the intellectual landscape, part of larger activities of editing and design. A crucial characteristic of collections is variety of philosophical urgency they respond to.

The electronic operates under the sign of limitless storage… potentially boundless inclusiveness. Design logic is a craving for elucidation, more context, the ability for the reader to follow any line of thought they might be having and follow it to the end. Unlimited informational desire, closing in of intellectual constraints. How do boundedness and internal cohesion help define the purpose of a collection? Deliberate attempt at genre not limited by technical limitations. Boundedness helps define and reflect philosophical purpose.

What do we model when we design and build digital collections? We're modelling the agency through which the collection comes into being and is sustained through usage. Design is a collection of representational practices, item selection, item boundaries and contents. There's a homogeneity in the structure, the markup applied to items. Item-to-item interconnections – there's the collection-level 'explicit phenomena' – the directly comparable metadata through which we establish cross-sectional views through the collection (eg by Dublin Core fields) which reveal things we already know about texts – authorship of an item, etc. There's also collection-level 'implicit phenomena' – informational commonalities, patterns that emerge or are revealed through inspection; change shape imperceptibly through how data is modelled or through software used [not sure I got that down right]; they're always motivated so always have a close connection with method.

Readerly knowledge – what can the collection assume about what the reader knows? A table of contents is only useful if you can recognise the thing you want to find in it – they're not always self-evident. How does the collection's modelling affect us as readers? Consider the effects of choices on the intellectual ecology of the collection, including its readers. Readerly knowledge has everything to do with what we think we're doing in digital humanities research.

The Hermeneutics of Screwing Around (pdf). Searching produces a dynamically located just-in-time collection… Search is an annoying guessing game with a passive-aggressive collection. But we prefer to ask a collection to show its hand in a useful way (i. e. browse)… Search -> browse -> explore.

What's the cultural significance of collections? She referenced Liu's Sidney's Technology… A network as flow of information via connection, perpetually ongoing contextualisation; a patchwork is understood as an assemblage, it implies a suturing together of things previously unrelated. A patchwork asserts connections by brute force. A network assumes that connections are there to be discovered, connected to. Patchwork, mosaic – connects pre-existing nodes that are acknowledged to be incommensurable.

We avow the desirability of the network, yet we're aware of the itch of edge cases, data that can't be brought under rule. What do we treat as noise and what as signal, what do we deny is the meaning of the collection? Is exceptionality or conformance to type the most significant case? On twitter, @aylewis summarised this as 'Patchworking metaphor lets us conceptualise non-conformance as signal not noise'

Pay attention to the friction in the system, rather than smoothing it over. Collections both express and support analysis. Expressing theories of genre etc in internal modelling… Patchwork – the collection articulates the scholarly interest that animated its creation but also interests of the reader… The collection is animated by agency, is modelled by it, even while it respects the agency we bring as readers. Scholarly enquiry is always a transaction involving agency on both ends.

My (not very good) notes from discussion afterwards… there was a question about digital femmage; discussion of the tension between the desire for transparency and the desire to permit many viewpoints on material while not disingenuously disavowing the roles in shaping the collection; the trend at one point for factoids rather than narratives (but people wanted the editors' view as a foundation for what they do with that material); the logic of the network – a collection as a set of parameters not as a set of items; Alan Liu's encouragement to continue with theme of human agency in understanding what collections are about (e.g. solo collectors like John Soane); crowdsourced work is important in itself regardless of whether it comes up with the 'best' outcome, by whatever metric. Flanders: 'the commitment to efficiency is worrisome to me, it puts product over people in our scale of moral assessment' [hoorah! IMO, engagement is as important as data in cultural heritage]; a question about the agency of objects, with the answer that digital surrogates are carriers of agency, the question is how to understand that in relation to object agency?

GIS and Mapping I

The first paper was 'Mapping the Past in the Present' by Andrew Wilson, which was a fast run-through some lovely examples based on Sydney's geo-spatial history. He discussed the spatial turn in history, and the mid-20thC shift to broader scales, territories of shared experience, the on-going concern with the description of space, its experience and management.

He referenced Deconstructing the map, Harley, 1989, 'cartography is seldom what the cartographers say it is'. All maps are lies. All maps have to be read, closely or distantly. He referenced Grace Karskens' On the rocks and discussed the reality of maps as evidence, an expression of European expansion; the creation of the maps is an exercise in power. Maps must be interpreted as evidence. He talked about deriving data from historic maps, using regressive analysis to go back in time through the sources. He also mentioned TGIS – time-enabled GIS. Space-time composite model – when have lots and lots of temporal changes, create polygon that describes every change in the sequence.

The second paper was 'Reading the Text, Walking the Terrain, Following the Map: Do We See the Same Landscape?' by Øyvind Eide. He said that viewing a document and seeing a landscape are often represented as similar activities… but seeing a landscape means moving around in it, being an active participant. Wood (2010) on the explosion of maps around 1500 – part of the development of the modern state. We look at older maps through modern eyes – maps weren't made for navigation but to establish the modern state.

He's done a case study on text v maps in Scandinavia, 1740s. What is lost in the process of converting text to maps? Context, vagueness, under-specification, negation, disjunction… It's a combination of too little and too much. Text has information that can't fit on a map and text that doesn't provide enough information to make a map. Under-specification is when a verbal text describes a spatial phenomenon in a way that can be understood in two different ways by a competent reader. How do you map a negative feature of a landscape? i.e. things that are stated not to be there. 'Or' cannot be expressed on a map… Different media, different experiences – each can mediate only certain aspects for total reality (Ellestrom 2010).

The third paper was 'Putting Harlem on the Map' by Stephen Robertson. This article on 'Writing History in the Digital Age' is probably a good reference point: Putting Harlem on the Map, the site is at Digital Harlem. The project sources were police files, newspapers, organisational archives… They were cultural historians, focussed on individual level data, events, what it was like to live in Harlem. It was one of first sites to employ geo-spatial web rather than GIS software. Information was extracted and summarised from primary sources, [but] it wasn't a digitisation project. They presented their own maps and analysis apart from the site to keep it clear for other people to do their work.  After assigning a geo-location it is then possible to compare it with other phenomena from the same space. They used sources that historians typically treat as ephemera such as society or sports pages as well as the news in newspapers.

He showed a great list of event types they've gotten from the data… Legal categories disaggregate crime so it appears more often in the list though was the minority of data. Location types also offers a picture of the community.

Creating visualisations of life in the neighbourhood…. when mapping at this detailed scale they were confronted with how vague most historical sources are and how they're related to other places. 'Historians are satisfied in most cases to say that a place is 'somewhere in Harlem'.' He talked about visualisations as 'asking, but not explaining, why there?'.

I tweeted that I'd gotten a lot more from his demonstration of the site than I had from looking at it unaided in the past, which lead to a discussion with @claudinec and @wragge about whether the 'search vs browse' accessibility issue applies to geospatial interfaces as well as text or images (i.e. what do you need to provide on the first screen to help people get into your data project) and about the need for as many hooks into interfaces as possible, including narratives as interfaces.

Crowdsourcing was raised during the questions at the end of the session, but I've forgotten who I was quoting when I tweeted, 'by marginalising crowdsourcing you're marginalising voices', on the other hand, 'memories are complicated'.  I added my own point of view, 'I think of crowdsourcing as open source history, sometimes that's living memory, sometimes it's research or digitisation'.  If anything, the conference confirmed my view that crowdsourcing in cultural heritage generally involves participating in the same processes as GLAM staff and humanists, and that it shouldn't be exploitative or rely on user experience tricks to get participants (though having made crowdsourcing games for museums, I obviously don't have a problem with making the process easier to participate in).

The final paper I saw was Paul Vetch, 'Beyond the Lowest Common Denominator: Designing Effective Digital Resources'. He discussed the design tensions between: users, audiences (and 'production values'); ubiquity and trends; experimentation (and failure); sustainability (and 'the deliverable'),

In the past digital humanities has compartmentalised groups of users in a way that's convenient but not necessarily valid. But funding pressure to serve wider audiences means anticipating lots of different needs. He said people make value judgements about the quality of a resource according to how it looks.

Ubiquity and trends: understanding what users already use; designing for intuition. Established heuristics for web design turn out to be completely at odds with how users behave.

Funding bodies expect deliverables, this conditions the way they design. It's difficult to combine: experimentation and high production values [something I've posted on before, but as Vetch said, people make value judgements about the quality of a resource according to how it looks so some polish is needed]; experimentation and sustainability…

Who are you designing for? Not the academic you're collaborating with, and it's not to create something that you as a developer would use. They're moving away from user testing at the end of a project to doing it during the project. [Hoorah!]

Ubiquity and trends – challenges include a very highly mediated environment; highly volatile and experimental… Trying to use established user conventions becomes stifling. (He called useit.com 'old nonsense'!) The ludic and experiential are increasingly important elements in how we present our research back.

Mapping Medieval Chester took technology designed for delivering contextual ads and used it to deliver information in context without changing perspective (i.e. without reloading the page, from memory).  The Gough map was an experiment in delivering a large image but also in making people smile.  Experimentation and failure… Online Chopin Variorum Edition was an experiment. How is the 'work' concept challenged by the Chopin sources? Technical methodological/objectives: superimposition; juxtaposition; collation/interpolation…

He discussed coping strategies for the Digital Humanities: accept and embrace the ephemerality of web-based interfaces; focus on process and experience – the underlying content is persistent even if the interfaces don't last.  I think this was a comment from the audience: 'if a digital resource doesn't last then it breaks the principle of citation – where does that leave scholarship?'

Summary

So those are my notes.  For further reference I've put a CSV archive of #DHA2012 tweets from searchhash.com here, but note it's not on Australian time so it needs transposing to match the session times.

This was my first proper big Digital Humanities conference, and I had a great time.  It probably helped that I'm an Australian expat so I knew a sprinkling of people and had a sense of where various institutions fitted in, but the crowd was also generally approachable and friendly.

I was also struck by the repetition of phrases like 'the digital deluge', the 'tsunami of data' – I had the feeling there's a barely managed anxiety about coping with all this data. And if that's how people at a digital humanities conference felt, how must less-digital humanists feel?

I was pleasantly surprised by how much digital history content there was, and even more pleasantly surprised by how many GLAMy people were there, and consequently how much the experience and role of museums, libraries and archives was reflected in the conversations.  This might not have been as obvious if you weren't on twitter – there was a bigger disconnect between the back channel and conversations in the room than I'm used to at museum conferences.

As I mentioned in my day 1 and day 2 posts, I was struck by the statement that 'history is on a different evolutionary branch of digital humanities to literary studies', partly because even though I started my PhD just over a year ago, I've felt the title will be outdated within a few years of graduation.  I can see myself being more comfortable describing my work as 'digital history' in future.

I have to finish by thanking all the speakers, the programme committee, and in particular, Dr Paul Arthur and Dr Katherine Bode, the organisers and the aaDH committee – the whole event went so smoothly you'd never know it was the first one!

And just because I loved this quote, one final tweet from @mikejonesmelb: Sir Ken Robinson: 'Technology is not technology if it was invented before you were born'.

Quick and dirty Digital Humanities Australasia notes: day 2

What better way to fill in stopover time in Abu Dhabi than continuing to post my notes from DHA2012? [Though I finished off the post and re-posted once I was back home.] These are my very rough notes from day 2 of the inaugural Australasian Association for Digital Humanities conference (see also Quick and dirty Digital Humanities Australasia notes: day 1 and Slow and still dirty Digital Humanities Australasia notes: day 3). In the interests of speed I'll share my notes and worry about my own interpretations later.

Keynote panel, 'Big Digital Humanities?'

Day 2 was introduced by Craig Bellamy, and began with a keynote panel with Peter Robinson, Harold Short and John Unsworth, chaired by Hugh Craig. [See also Snurb's liveblogs for Robinson, Short and Unsworth.] Robinson asked 'what constitutes success for the digital humanities?' and further, what does the visible successes of digital humanities mask? He said it's harder for scholars to do high quality research with digital methods now than it was 20 years ago. But the answer isn't more digital humanists, it's having the ingredients to allow anyone to build bridges… He called for a new generation of tools and methods to support the scholarship that people want to do: 'It should be as easy to make a digital edition (of a document/book) as it is to make a Facebook page', it shouldn't require collaboration with a digital humanist. To allow data made by one person to be made available to others, all digital scholarship should be made available under a Creative Commons licence (publishers can't publish it now if it's under a non-commercial licence), and digital humanities data should be structured and enriched with metadata and made available for re-use with other tools. The model for sustainability depends on anyone and everyone being able to access data.

Harold Short talked about big (or at least unescapable) data and the 'Svensson challenge' – rather than trying to work out how to take advantage of infrastructure created by and for the sciences, use your imagination to figure out what's needed for the arts and humanities. He called for a focus on infrastructure and content rather than 'data'.

John Unsworth reminded us that digital humanities is a certain kind of work in the humanities that uses computational methods as its research methods. It's not just using digital materials, though it does require large collections of data – it also requires a sense of how how the tools work.

What is the digital humanities?

Very different versions of 'digital humanities' emerged through the panel and subsequent discussion, leaving me wondering how they related to the different evolutionary paths of digital history and digital literature studies mentioned the day before. Meanwhile, on the back channel (from the tweets that are to hand), I wondered if a two-tier model of digital humanities was emerging – one that uses traditional methods with digital content (DH lite?); another that disrupts traditional methods and values. Though thinking about it now, the 'tsunami' of data mentioned is disruptive in its own right, regardless of the intentional choices one makes about research practices (which might have been what Alan Liu meant when he asked about 'seamless' and 'seamful' views of the world)…. On twitter, other people (@mikejonesmelb, @bestqualitycrab, @1n9r1d) wondered if the panel's interpretation of 'big' data was gendered, generational, sectoral, or any other combination of factors (including as the messiness and variability of historical data compared to literature) and whether it could have been about 'disciplinary breadth and inclusiveness' rather than scale.

Data morning session

The first speaker was Toby Burrows on 'Using Linked Data to Build Large‐Scale e‐Research Environments for the Humanities'. [Update: he's shared his slides and paper online and see also Snurb's liveblog.] Continuing some of the themes from the morning keynote panel, he said that the humanities has already been washed away in the digital deluge, the proliferation of digital stuff is beyond the capacity of individual researchers. It's difficult to answer complex humanities questions only using search with this 'industrialised' humanities data, but large-scale digital libraries and collections offer very little support for functions other than search. There's very little connection between data that researchers are amassing and what institutions are amassing.

He's also been looking at historians/humanists research practices [and selfishly I was glad to see many parallels with my own early findings]. The tools may be digital rather than paper and scissors, but historians are still annotating and excerpting as they always have. The 'sharing' part of their work has changed the most – it's easier to share, and they can share at an earlier stage if they choose to do that, but not a lot has changed at the personal level.

Burrows said applying applying linked data approach to manuscript research would go a long way to addressing the complexity of the field. For example, using global URIs for manuscripts and parts; separating names and concepts from descriptive information; and using linked data functions to relate scholarly activities (annotations, excerpts, representations etc) to manuscript descriptions, objects and publications. Linked data can provide a layer of entities that sits between research activities and descriptions/collections/publications, which avoids conflating the entities and the source material. Multiple naming schemes are necessary for describing entities and relationships – there's no single authoritative vocabulary. It's a permanent work in progress, with no definitive or final structure. Entities need to include individuals as well as categories, with a network graph showing relatedness and the evidence for that relatedness as the basic structure.

He suggested a focus on organising knowledge, not collections, whether objects or texts. Collaborative activities should be based around this knowledge, using tools that work with linked data entities. This raised the issue of contested ground and the application of labels and meaning to data: your 'discovery' is my 'invasion'. This makes citizen humanities problematic – who gets to describe, assign, link, and what does that mean for scholarly authority?

My notes aren't clear but I think Burrows said these ideas were based on analysis of medieval manuscript research, which Jane Hunter had also worked on, and they were looking towards the architecture for HuNI. It was encouraging to see an approach to linked data so grounded in the complexity of historians research practices and data, and is yet another reason I'm looking forward to following HuNI's progress – I think it will have valuable lessons for linked data projects in the rest of the world. [These slides from the Linked Open Data workshop in Melbourne a few weeks later show the academic workflow HuNI plans to support and some of the issues they'll have to tackle.]

The second speaker was the University of Sydney's Stephen Hayes on 'how linked is linked enough?'. [See also Snurb's liveblog.] He's looking at projects through a linked data lens, trying to assess how much further projects need to go to comfortably claim to be linked data. He talked about the issues projects encountered trying to get to be 5 star Linked Data.

He looked at projects like the Dictionary of Sydney, which expresses data as RDF as well in a public-facing HTML interface and comes close to winning 5 stars. It is a demonstration of the fact that once data is expressed in one form, it can be easily expressed in another form – stable entities can be recombined to form new structures. The project is powered by Heurist, a tool for managing a wide range of research data. The History of Balinese Painting could not find other institutions that exposed Balinese collection data in programmable form so they could link to them (presumably a common problem for early adopters but at least it helps solve the 'chicken or the egg' problem that dogs linked data in cultural heritage and the humanities). The sites URLs don't return useful metadata but they do try to refer to image URLs so it's 'sorta persistent'. He gave it a rating of 3.5 stars. Other projects mentioned (also built on Heurist?) were the Charles Harpur Critical Archive, rated at 3.5 stars and Virtual Zagora, rated at 3 stars.

The paper was an interesting discussion of the team work required to get the full 5 stars of linked data, and the trade-offs in developing functions for structured data (e.g. implementing schema.org's painting markup versus focussing on the quality of the human-facing pages); reassuring curators about how much data would be released and what would be kept back; developing ontologies throughout a project or in advance and the overhead in mapping other projects concepts to their own version of Dublin Core.

The final paper in the session was 'As Curious An Entity: Building Digital Resources from Context, Records and Data' by Michael Jones and Antonina Lewis (abstract). [See also Snurb's liveblog.] They said that improving the visibility of relationships between entities enriches archives, as does improving relationships between people. The title quote in full is 'as curious an entity as bullshit writ on silk' – if the parameters, variables and sources of data are removed from material, then it's just bullshit written on silk. Visualisations remove sources, complexity and 'relative context', and would be richer if they could express changes in data over time and space. They asked how one would know that information presented in a visualisation is accurate if it doesn't cite sources? You must seek and reference original material to support context layers.

They presented an overview of the Saulwick Archive project (Saulwick ran polls for the Fairfax newspapers for years) and the Australian Women's Register, discussed common issues faced in digital humanities, and the role of linked data and human relationships in building digital resources. They discussed the value of maintaining relationships between archives and donors after the transfer of material, and the need to establish data management plans to make provision for raw data and authoritative versions of related contextual material, and to retain data to make sense of the archives in the future. The Australian Women's Register includes content written for the site and links out to the archival repositories and libraries where the records are held. In a lovely phrase, they described records as the 'evidential heart' for the context and data layers. They also noted that the keynote overlooked non-academic re-use of digital resources, but it's another argument for making data available where possible.

Digital histories session

The first paper was 'Community Connections: The Renaissance of Local History' by Lisa Murray. Murray discussed the 'three Cs' needed for local history: connectivity, community, collaboration.

Is the process of geo-referencing forcing historians to be more specific about when or where things happened? Are people going from the thematic to the particular? Is it exciting for local historians to see how things fit into state or national narratives? Digital history has enormous potential for local and family history and to represent complicated relationships within a community and how they've changed over time. Digital history doesn't have to be article-centric – it enables new forms of presentation. Historians have to acknowledge that Wikipedia is aligned to historians' processes. Local history is strongly represented on Wikipedia. The Dictionary of Sydney provides a universal framework for accessing Sydney's history.

The democratisation of historical production is exciting but raises it challenges for public understandings of how history undertaken and represented. Are some histories privileged? Making History (a project by Museum Victoria and Monash University) encourages the use of online resources but does that privilege digitised sources, and will others be neglected? Are easily accessible sources privileged, and does that change what history is written? What about community collections or vast state archives that aren't digitised?

History research methodologies are changing – Google etc is shaping how research is undertaken; the ubiquity of keyword searching reinforces the primacy of names. She noted the impact of family historians on how archives prioritise work. It's not just about finding sources – to produce good history you need to analyse the sources. Professional historians are no longer the privileged producers of knowledge. History can be parochial, inclusive, but it can also lack sense of historical perspective, context. Digital history production amplifies tensions between popular history and academic history [and presumably between amateur and academic historians?].

Apparently primary school students study more local history than university students do. Local and community history is produced by broad spectrum of community but relatively few academic historians are participating. There's a risk of favouring quirky facts over significance and context. Unless history is more widely taught, local history will be tarred with same brush as antiquarians. History is not only about narrative and context… Historians need to embrace the renaissance of local and community history.

In the questions there was some discussion of the implications of Sydney's city archives being moved to a more inconvenient physical location. The justification is that it's available through Ancestry but that removes it from all context [and I guess raises all the issues of serendipity etc in digital vs physical access to archives].

The next speaker was Tim Sherratt on 'Inside the bureaucracy of White Australia'. His slides are online and his abstract is on the Invisible Australians site. The Invisible Australians project is trying to answer the question of what the White Australia policy looked like to a non-white Australian.  He talked about how digital technology can help explore the practice of exclusion as legislation and administrative processes were gradually elaborated. Chinese Australians who left Australia and wanted to return had to prove both their identity and their right to land to convince officials they could return: 'every non-white resident was potentially a prohibited immigrant just waiting to be exposed'. He used topic modelling on file titles from archival series and was able to see which documents related to the White Australia policy. This is a change from working through hierarchical structures of archives to working directly through the content of archives. This provides a better picture of what hasn't survived, what's missing and would have many other exciting uses. [His post on Topic modelling in the archives explains it better than my summary would.]

The final paper was Paul Turnbull on 'Pancake history'. He noted that in e-research there's a difference between what you can use in teaching and what makes people nervous in the research domain. He finds it ironic that professional advancement for historians is tied to writing about doing history rather than doing history. He talked about the need to engage with disciplinary colleagues who don't engage with digital humanities, and issues around historians taking digital history seriously.

Sherratt's talk inspired discussion of funding small-scale as well as large-scale infrastructure, possibly through crowdfunding. Turnbull also suggested 'seeding ideas and sharing small apps is the way to go'.

[Note from when I originally posted this: I don't know when my flight is going to be called, so I'll hit publish now and keep working until I board – there's lots more to fit in for day 2! In the afternoon I went to the 'Digital History' session. I'll tidy up when I'm in the UK as I think blogger is doing weird LTR things because it may be expecting Arabic.]

See also Slow and still dirty Digital Humanities Australasia notes: day 3.

Quick and dirty Digital Humanities Australasia notes: day 1

As always, I should have done this sooner and tidied them up more, but better rough notes than nothing, so here goes… The Australasian Association for Digital Humanities held their inaugural conference in Canberra in March, 2012.  You can get an overall sense of the conference from the #DHA2012 tweets (I've put a CSV archive of #DHA2012 tweets from searchhash.com here, but note it's not on Australian time) and from the keynotes.

In his opening keynote on the movements between close and distant reading, Alan Liu observed that the crux of the 'reading' issue depends on the field, and further, that 'history is on a different evolutionary branch of digital humanities to literary studies'.  This is something I've been wondering about since finding myself back in digital humanities, and was possibly reflected in the variety of papers in the overall programme.  I was generally following sessions on digital history, geospatial themes and crowdsourcing, but there was so much in the programme that you could have followed a literary studies line and had a totally different conference experience.

In the next session I went to a panel on 'Connecting Australia's Cultural Datasets: A Vision for Collaboration' with various people from the new 'Humanities Networked Infrastructure' (HuNI) (more background) presenting.  It started with Deb Verhoeven on 'jailbreaking cultural data' and the tension identified by Brand: "information wants to be expensive because it's so valuable.  The right information in the right place just changes your life.  On the other hand, information wants to be free, because the cost of getting it out is lower and lower all the time. So you have these two things fighting against each other". 'Information wants to be social': she discussed the need to understand the value of research in terms of community engagement, not just as academically ranked output, and to return research to the communities they're investigating in meaningful ways.
 
Other statements that resonated were the need for organisational, semantic and technical interoperability in datasets to create collaborative environments. Collaboration requires data integration and exchange as well as dealing with different ideas about what 'data' is in different disciplines in the humanities. Collaboration in the cultural datasets community can follow unmet needs: discover data that's currently hidden, make connections between disparate data sources, publish and share connections.

Ross Harley talked about how interoperability facilitates serendipity and trying to find new ways for data to collide. In the questions, Ingrid Mason asked about parallels with the GLAM (galleries, libraries, archives and museums) community, but it was also pointed out that GLAMs are behind in publishing their data – not everything HuNI wants to use is available yet.  I pointed out (on the twitter back channel) that requests for GLAM information from intensive users (e.g. researchers) helps memory institutions make the case for publishing more data – it's still all a bit chicken-or-the-egg.

After lunch I went to the crowdsourcing session (not least cos I was presenting early results from my PhD in it).  The first presentation was on 'crowdsourcing semantic tags on 3D museum artefacts' which could have amazing applications for teaching material culture and criticism as well as source communities because it lets people annotate specific locations on a 3D model. Interestingly, during the questions someone reported people visiting campus classics museum who said they were enjoying seeing the objects in person but also wanted access to electronic versions – it's fascinating watching audience expectations change.

The next presentation was on 'Optimising crowdsourcing websites to increase volunteer participation' which was a case study of NYPL's What's on the menu by Donelle McKinley who was using MECLAB/Flint McGlaughlin's Conversion Sequence heuristic (clarity of value proposition, motivation, incentive, friction, anxiety) to assess how the project's design was optimised to motivate audience participation.  Donelle's analysis is really useful for people thinking about designing for crowdsourcing, but I'm not sure my notes do it justice, and I'm afraid I didn't get many notes for Pauline Cockrill's 'Using Web 2.0 to make new connections in community history' as I was on just afterwards.  One point I tweeted was about a quick win for crowdsourcing in using real-world communities as pointers to successful online collaborations, but I'm not sure now who said it.

One comment I noted during the discussion was "a real pain about Old Weather was that you'd get into working on a ship and it would just sail off on you" – interfaces that work for the organisation doesn't always work for the audience.  This session was generally useful for clarifying my thoughts on the tension between optimising for efficiency or engagement in cultural heritage crowdsourcing projects.

In the interests of getting this posted I'll stop here and call this 'day 1'. I'm not sure if any of the slides are available yet, but I'll update and link to any presentations or other write-ups I find. There's a live blog of many sessions at http://snurb.info/taxonomy/term/137.

[Update: I've posted about Day 2 at Quick and dirty Digital Humanities Australasia notes: day 2 and Slow and still dirty Digital Humanities Australasia notes: day 3.]

How things change: the Google Art Project (again)

The updated Google Art Project has been launched with loads more museums contributing over 30,000 artworks.  The interface still seems a bit sketchy to me (sometimes you can open links in a new tab, sometimes you can't; mystery meat navigation; the lovely zoom option isn't immediately discoverable; the thumbnails that appear at the bottom don't have a strong visual connection with the action that triggers their appearance; and the only way I could glean any artist/title information about the thumbnails was by looking at the URL), but it's nice to see options for exploring by collection (collecting institution, I assume), date or artist emphasised in the interface. 

Anyway, it's all about the content – easy access to high-quality zoomable images of some of the world's best artworks in an interface with lots of relevant information and links back to the holding institution is a win for everyone.  And if the attention (and traffic) makes museums a little jealous, well, it'll be fascinating to see how that translates into action.  After all, keeping up with the Joneses seems to be one way museums change…

Reading some online stories about the launch, I was struck by how far conversations about traditional and online galleries have come.  From one:

As users explore the galleries they can also add comments to each painting and share the whole collection with friends and family. Try doing that in the Tate Modern. Actually, don’t.

Although, of course, you can – it's traditionally known as 'having a conversation in a museum'. 
But in 2012, is visiting a website and sharing links online seen as a reasonable stand-in for the physical visit to a museum, leaving the in-person gallery visit for 'purists' and enthusiasts?  (This might make blockbuster exhibtions bearable.)  Or, as the consensus of the past decade has it, does it just whet the appetite and create demand for an experience with the original object, leading to more visits?

Can you capture visitors with a steampunk arm?

Credits: Science Museum

This may be familiar to you if you've worked on a museum website: an object will capture the imagination of someone who starts to spread the link around, there's a flurry of tweets and tumblrs and links (that hopefully you'll notice in time because you've previously set up alerts for keywords or URLs on various media), others like it too and it starts to go viral and 50,000 people look at that one page in a day, 20,000 the next, furious discussions break out on social media and other sites… then they're gone, onto the next random link on someone else's site.  It's hugely exciting, but it can also feel like a missed opportunity to show these visitors other cool things you have in your collection, to address some of the issues raised and to give them more information about the object.

There are three key aspects to riding these waves of interest: the ability to spot content that's suddenly getting a lot of hits; the ability to respond with interesting, relevant content while the link is still hot (i.e. within anything from a couple of hours to a couple of days); and the ability to put that relevant content on the page where fly-by-night visitors will see it.

For many museums, caught between a templated CMS and layers of sign-off for new content , it's not as easy as it sounds.  When the Science Museum's 'steampunk artificial arm' started circulating on twitter and then made boingboing, I was able to work with curators to get a post on the collections blog about it the next day, but then there was no way of adding that link to the Brought to Life page that was all most people saw.

In his post on “The Guardian’s Facebook app”, Martin Belam discusses how their Facebook app has helped archived content live again:

Someone shares an old article with their friends, some of their friends either already use or install the app, and the viral effect begins to take hold. … We’ve got over 1.3 million articles live on the website, so that is a lot of content to be discovered, and the app means that suddenly any page, languishing unloved in our database, can become a new landing page. When an article becomes popular in the app, we sometimes package it with content. Because we know the attention has come at a specific time from a specific place, we can add related links that are appropriate to the audience rather than to the original content. …when you’ve got the audience there, you need to optimise for them

As a content company with great technical and user experience teams, the Guardian is better placed to put together existing content around a viral article, but still, I'm curious: are any museums currently managing to respond to sudden waves of interest in random objects?  And if so, how?

Notes from Culture Hack Day (#chd11)

Culture Hack Day (#chd11) was organised by the Royal Opera House (the team being @rachelcoldicutt, @katybeale, @beyongolia, @mildlydiverting, @dracos – and congratulations to them all on an excellent event). As well as a hack event running over two days, they had a session of five minute 'lightning talks' on Saturday, with generous time for discussion between sessions. This worked quite well for providing an entry point to the event for the non-technical, and some interesting discussion resulted from it. My notes are particularly rough this time as I have one arm in a sling and typing my hand-written notes is slow.

Lightning Talks
Tom Uglow @tomux “What if the Web is a Fad?”
'We're good at managing data but not yet good at turning it into things that are more than points of data.' The future is about physical world, making things real and touchable.

Clare Reddington, @clarered, “What if We Forget about Screens and Make Real Things?”
Some ace examples of real things: Dream Director; Nuage Vert (Helsinki power station projected power consumption of city onto smoke from station – changed people's behaviour through ambient augmentation of the city); Tweeture (a conch, 'permission object' designed to get people looking up from their screens, start conversations); National Vending Machine from Dutch museum.

Leila Johnston, @finalbullet talked about why the world is already fun, and looking at the world with fresh eyes. Chromaroma made Oyster cards into toys, playing with our digital footprint.

Discussion kicked off by Simon Jenkins about helping people get it (benefits of open data etc) – CR – it's about organisational change, fears about transparency, directors don't come to events like this. Understand what's meant by value – cultural and social as well as economic. Don't forget audiences, it has to be meaningful for the people we're making it (cultural products) for'.

Comment from @fidotheCultural heritage orgs have been screwed over by software companies. There's a disconnect between beautiful hacks around the edges and things that make people's lives easier. [Yes! People who work in cultural heritage orgs often have to deal with clunky tools, difficult or vendor-dependent data export proccesses, agencies that over-promise and under-deliver. In my experience, cultural orgs don't usually have internal skills for scoping and procuring software or selecting agencies so of course they get screwed over.]

TU: desire to be tangible is becoming more prevalent, data to enhance human experience, the relationship between culture and the way we live our lives.

CR: don't spend the rest of the afternoon reinforcing silos, shouldn't be a dichotomy between cultural heritage people and technologists. [Quick plug for http://museum30.ning.com/, http://groups.google.com/group/antiquist, http://museum-api.pbwiki.com/ and http://museumscomputergroup.org.uk/email-list/ as places where people interested in intersection between cultural heritage and technology can mingle – please let me know of any others!] Mutual respect is required.

Tom Armitage, @infovore “Sod big data and mashups: why not hack on making art?”
Making culture is more important than using it. 3 trends: 1) collection – tools to slice and dice across time or themes; 2) magic materials 3) mechanical art, displays the shape of the original content; 3a) satire – @kanyejordan 'a joke so good a machine could make it'.

Tom Dunbar, @willyouhelp – story-telling possibilites of metadata embedded in media e.g. video [check out Waisda? for game designed to get metdata added to audio-visual archives]. Metadata could be actors, characters, props, action…

Discussion [?]:remixing in itself isn't always interesting. Skillful appropriation across formats… Universe of editors, filterers, not only creators. 'in editing you end up making new things'.

Matthew Somerville, @dracos, Theatricalia, “What if You Never Needed to Miss a Show?”
'Quite selfish', makes things he needs. Wants not to miss theatre productions with people he likes in/working on them. Theatricalia also collects stories about productions. [But in discussion it came up that the National Theatre asked him to remove data – why?! A recommendation system would definitely get me seeing more theatre, and I say that as a fairly regular but uninformed theatre-goer who relies on word-of-mouth to decide where to spend ticket money.]

Nick Harkaway, @Harkaway on IP and privacy
IP as way of ringfencing intangible ideas, requiing consent to use. Privacy is the same. Not exciting, kind of annoying but need to find ways to make it work more smoothly while still proving protection. 'Buying is voting', if you buy from Tesco, you are endorsing their policies. 'Code for the change you want to see in the world', build the tools you want cultural orgs to have so they can do better. [Update: Nick has posted his own notes at Notes from Culture Hack Day. I really liked the way he brought ethical considerations to hack enthusiasm for pushing the boundaries of what's possible – the ability to say 'no' is important even if a pain for others.]

Chris Thorpe, @jaggeree. ArtFinder, “What if you could see through the walls of every museum and something could tell you if you’d like it?”

Culture for people who don't know much about culture. Cultural buildings obscure the content inside, stop people being surprised by what's available. It's hard if you don't know where to start. Go for user-centric information. Government Art Collection Explorer – ace! Wants an angel for art galleries to whisper information about the art in his ear. Wants people to look at the art, not the screen of their device [museums also have this concern]. SAP – situated audio platform. Wants a 'flight data recorder' for trips around cultural places.

Discussion around causes of fear and resistance to open data – what do cultural orgs fear and how can they learn more and relax? Fear of loss of provenance – response was that for developers displaying provenance alongside the data gives it credibility; counter-response was that organisations don't realise that's possible. [My view is that the easiest way to get this to change is to change the metrics by which cultural heritage organisations are judged, and resolve the tension between demands to commercialise content to supplement government grants and demands for open access to that same data. Many museums have developed hybrid 'free tombstone, low-res, paid-for high-res' models to deal with this, but it's taken years of negotiation in each institution.] I also ranted about some of these issues at OpenTech 2010, notes at 'Museums meet the 21st century'.

Other discussion and notes from twitter – re soap/drama characters tweeting – I managed to out myself as a Neighbours watcher but it was worth it to share that Neighbours characters tweet and use Facebook. Facebook relationship status updates and events have been included as plot points, and references are made to twitter but not to the accounts of the characters active on the service. I wonder if it's script writers or marketing people who write the characters tweets? They also tweet in sync with the Australian showings, which raises issues around spoilers and international viewers.

Someone said 'people don't want to interact with cultural institutions online. They want to interact with their content' but I think that's really dependent on the definition of content – as pointed out, points of data have limited utility without further context. There's a catch-22 between cultural orgs not yet making really engaging data and audiences not yet demanding it, hopefully hack days like CHD11 help bridge the gap and turn data into stories and other meaningful content. We're coming up against the limits of what can be dome programmatically, especially given variation in quality and extent of cultural heritage data (and most of it is data rather than content).

[Update: after writing this I found a post The lightning talks at Culture Hack Day about the day, which happily picks up on lots of bits I missed. Oh, and another, by Roo Reynolds.]

After the lightning talks I popped over the road to check out the hacking and ended up getting sucked in (the lure of free pizza had a powerful effect!).  I worked on a WordPress plugin with Ian Ibbotson @ianibbo that lets you search for a term on the Culture Grid repository and imports the resulting objects into my museum metadata games so that you can play with objects based on your favourite topic.  I've put the code on github [https://github.com/mialondon/mmg-import] and will move it from my staging server to live over the next few days so people can play with the objects.  It's such a pain only having one hand, and I'm very grateful to Ian for the chance to work together and actually get some code written.  This work means that any organisation that's contributed records to the Culture Grid can start to get back tags or facts to enhance their collections, based on data generated by people playing the games.  The current 300-ish objects have about 4400 tags and 30 facts, so that's not bad for a freebie. OTOH, I don't know of many museums with the ability to display content created by others on their collections pages or store it in their collections management systems – something for another hack day?

Something I think I'll play around with a bit more is the idea of giving cultural heritage data a quality rating as it's ingested.  We discussed whether the ratings would be local to an app (as they could be based on the particular requirements of that application) or generalised and recorded in the CultureGrid service.  You could record the provence of a rating which might be an approach that combines the benefits of both approaches.  At the moment, my requirements for a 'high quality' record would be: title (e.g. 'The Ashes trophy', if the object has one), name or type of object (e.g. cup), date, place, decent sized image, description.

Finally, if you're interested in hacking around cultural heritage data, there's also historyhackday next weekend. I'm hoping to pop in (dependent on fracture and MSc dissertation), not least because in March I'm starting a PhD in digital humanities, looking at participatory digitisation of geo-located historical material (i.e. getting people to share the transcriptions and other snippets of ad hoc digitisation they do as part of their research) and it's all hugely relevant.

Notes on 'User Generated Content' session, Open Culture Conference 2010

My notes from the 'user generated content' parallel track on first day of the Open Culture 2010 conference. The session started with brief presentations by panellists, then group discussions at various tables on questions suggested by the organisers. These notes are quite rough, and of course any mistakes are mine. I haven't had a chance to look for the speakers' slides yet so inevitably some bits are missing, and I can only report the discussion at the table I was at in the break-out session. I've also blogged my notes from the plenary session of the Open Culture 2010 conference.

User-generated content session, Open Culture, Europeana – the benefits and challenges of UGC.
Kevin Sumption, User-generated content, a MUST DO for cultural institutions
His background – originally a curator of computer sciences. One of first projects he worked on at Powerhouse was D*Hub which presented design collections from V&A, Brooklyn Museum and Powerhouse Museum – it was for curators but also for general public with an interest in design. Been the source of innovation. Editorial crowd-sourcing approach and social tagging, about 8 years ago.

Two years ago he moved to National Maritime Museum, Royal Observatory, Greenwich. One of the first things they did was get involved with Flickr Commons – get historic photographs into public domain, get people involved in tagging. c1000 records in there. General public have been able to identify some images as Adam Villiers images – specialists help provide attribution for the photographer. Only for tens of records of the 000s but was a good introduction to power of UGC.

Building hybrid exhibition experiences – astronomy photographer of the year – competition on Flickr with real world exhibition for the winners of the competition. 'Blog' with 2000 amateur astronomers, 50 posts a day. Through power of Flickr has become a significant competition and brand in two years.

Joined citizen science consortia. Galaxy Zoo. Brainchild of Oxford – getting public engaged with real science online. Solar Stormwatch c 3000 people analysing and using the data. Many people who get involved gave up science in high school… but people are getting re-engaged with science *and* making meaningful contributions.

Old Weather – helping solve real-world problems with crowdsourcing. Launched two months ago.
Passion for UGC is based around where projects can join very carefully considered consortia, bringing historical datasets with real scientific problems. Can bring large interested public to the project. Many of the public are reconnecting with historical subject matter or sciences.

Judith Bensa-Moortgat, Nationaal Archief, Netherlands, Images for the Future project
Photo collection of more than 1 million photos. Images for the future project aims to save audio-visual heritage through digitisation and conservation of 1.2 million photos.

Once digitised, they optimise by adding metadata and context. Have own documentalists who can add metadata, but it would take years to go through it all. So decided to try using online community to help enrich photo collections. Using existing platforms like Wikipedia, Flickr, Open Street map, they aim to retrieve contextual info generated by the communities.  They donated political portraits to Wikimedia Commons and within three weeks more than half had been linked to relevant articles.

Their experiences with Flickr Commons – they joined in 2008. Main goal was to see if community would enrich their photos with comments and tags. In two weeks, they had 400,000 page views for 400 photos, including peaks when on Dutch TV news. In six months, they had 800 photos with over 1 million views. In Oct 2010, they are averaging 100,000 page views a month; 3 million overall.

But what about comments etc? Divided them into categories of comments [with percentage of overall contributions]:

  • factual info about location, period, people 5%; 
  • link to other sources eg Wikipedia 5%; 
  • personal stories/memories (e.g. someone in image was recognised); 
  • moral discussions; 
  • aesthetical discussions; 
  • translations.

The first two are most important for them.
13,000 tags in many languages (unique tags or total?).
10% of the contributed UGC was useful for contextualisation; tags ensure accessibility [discoverability?] on the web; increased (international) visibility. [Obviously the figures will vary for different projects, depending on what the original intent of the project was]

The issues she'd like to discuss are – copyright, moderation, platforms, community.

Mette Bom, 1001 Stories about Denmark
Story of the day is one of the 1001 stories. It's a website about the history and culture of Denmark. The stories have themes, are connected to a timeline.  Started with 50 themes, 180 expert writers writing the 1001 stories, now it's up to the public to comment and write their own stories. Broad definition of what heritage is – from oldest settlement to the 'porn street' – they wanted to expand the definition of heritage.

Target audiences – tourists going to those places; local dedicated experts who have knowledge to contribute. Wanted to take Danish heritage out of museums.

They've created the main website, mobile apps, widget for other sites, web service.  Launched in May 2010.  20,000 monthly users. 147 new places added, 1500 pictures added.

Main challenges – how to keep users coming back? 85% new, 15% repeat visitors (ok as aimed at tourists but would like more comments). How to keep press interested and get media coverage? Had a good buzz at the start cos of the celebrities. How to define participation? Is it enough to just be a visitor?

Johan Oomen, Netherlands Institute for Sound and Vision, Vrij Uni Amsterdam. Participatory Heritage: the case of the Waisda? video labelling game.
They're using game mechanisms to get people to help them catalogue content. [sounds familiar!]
'In the end, the crowd still rules'.
. Tagging is a good way to facilitate time-based annotation [i.e. tag what's on the screen at different times]

Goal of game is consensus between players. Best example in heritage is steve.museum; much of the thinking about using tagging as a game came from Games with a Purpose (gwap.com).  Basic rule – players score points when their tag exactly matches the tag entered by another within 10 seconds. Other scoring mechanisms.  Lots of channels with images continuously playing.

Linking it to twitter – shout out to friends to come join them playing.  Generating traffic – one of the main challenges. Altruistic message 'help the archive' 'improve access to collections' came out of research with users on messages that worked. Worked with existing communities.

Results, first six months – 44,362 pageviews. 340,000 tags to 604 items, 42,068 unique tags.
Matches – 42% of tags entered more than 2 times. Also looked at vocab (GTAA, Cornetto), 1/3 words were valid Dutch words, but only a few part of thesauruses.  Tags evaluated by documentalists. Documentary film 85% – tags were useful; for reality series (with less semantic density) tags less useful.

Now looking at how to present tags on the catalogue Powerhouse Museum style.  Experimenting with visualising terms, tag clouds when terms represented, also makes it easy to navigate within the video – would have been difficult to do with professional metadata.  Looking at 'tag gardening' – invite people to go back to their tags and click to confirm – e.g. show images with particular tags, get more points for doing it.

Future work – tag matching – synonyms and more specific terms – will get more points for more specific terms.

Panel overview by Costis Dallas, research fellow at Athena, assistant professor at Panteion University, Athens.
He wants to add a different dimension – user-generated content as it becomes an object for memory organisations. New body of resources emerging through these communication practices.
Also, we don't have a historiography anymore; memory resides in personal information devices.  Mashups, changes in information forms, complex composed information on social networks – these raise new problems for collecting – structural, legal, preservation in context, layered composition.  What do we need to do now in order to be able to make use of digital technologies in appropriate, meaningful ways in the future? New kinds of content, participatory curation are challenges for preservation.

Group discussion (breakout tables)
Discussion about how to attract users. [It wasn't defined whether it was how to attract specifically users who'll contribute content or just generally grow the audience and therefore grow the number of content creators within the usual proportions of levels of participation e.g. Nielsen, Forrester; I would also have liked to discussed how to encourage particular kinds of contributions, or to build architectures of participation that provided positive feedback to encourage deeper levels of participation.]

Discussion and conclusions included – go with the strengths of your collections e.g. if one particular audience or content-attracting theme emerges, go with it.  Norway has a national portal where people can add content. They held lots of workshops for possible content creators; made contact with specialist organisations [from which you can take the lesson that UGC doesn't happen in a vacuum, and that it helps to invest time and resources into enabling participants and soliciting content].  Recording living history.  Physical presence in gallery, at events, is important.  Go where audiences already are; use existing platforms.

Discussion about moderation included – once you have comments, how are they integrated back into collections and digital asset management systems?  What do you do about incorrect UGC displayed on a page?  Not an issue if you separate UGC from museum/authoritative content in the interface design.  In the discussion it turned out that Europeana doesn't have a definition of 'moderation'.  IMO, it should include community management, including acknowledging and thanking people for contributions (or rather, moderation is a subset of community management).  It also includes approving or reviewing and publishing content, dealing with corrections suggested by contributors, dealing with incorrect or offensive UGC, adding improved metadata back to collections repositories.

User-generated content and trust – British Library apparently has 'trusted communities' on their audio content – academic communities (by domain name?) and 'everyone else'.  Let other people report content to help weed out bad content.

Then we got onto a really interesting discussion of which country or culture's version of 'offensive' would be used in moderating content.  Having worked in the UK and the Netherlands, I know that what's considered a really rude swear word and what's common vocabulary is quite different in each country… but would there be any content left if you considered the lowest common standards for each country?  [Though thinking about it later, people manage to watch films and TV and popular music from other countries so I guess they can deal with different standards when it's in context.]  To take an extreme content example, a Nazi uniform as memorabilia is illegal in Germany (IIRC) but in the UK it's a fancy dress outfit for a member of the royal family.

Panel reporting back from various table discussions
Kevin's report – discussion varied but similar themes across the two tables. One – focus on the call to action, why should people participate, what's the motivation? How to encourage people to participate? Competitions suggested as one solution, media interest (especially sustained). Notion of core group who'll energise others. Small groups of highly motivated individuals and groups who can act as catalysts [how to recruit, reward, retain]. Use social media to help launch project.

1001 Danish Stories promotional video effectively showed how easy the process of contributing content was,  and that it doesn't have to to be perfect (the video includes celebrities working the camera [and also being a bit daggy, which I later realised was quite powerful – they weren't cool and aloof]).
Giving users something back – it's not a one-way process. Recognition is important. Immediacy too – if participating in a project, people want to see their contributions acknowledged quickly. Long approval processes lose people.
Removal of content – when different social, political backgrounds with different notions of censorship.

Mette's report – how to get users to contribute – answers mostly to take away the boundaries, give the users more credit than we otherwise tend to. We always think users will mess things up and experts will be embarrassed by user content but not the case. In 1001 they had experts correcting other experts. Trust users more, involve experts, ask users what they want. Show you appreciate users, have a dialouge, create community. Make it a part of life and environment of users. Find out who your users are.

Second group – how Europeana can use the content provided in all its forms. Could build web services to present content from different places, linking between different applications.
How to set up goals for user activity – didn't get a lot of answers but one possibility is to start and see how users contribute as you go along. [I also think you shouldn't be experimenting with UGC without some goal in mind – how else will you know if your experiment succeeded?  It also focusses your interaction and interface design and gives the user some parameters (much more useful than an intimidating blank page)].

Judith's report (including our table) – motivation and moderation in relation to Europeana – challenging as Europeana are not the owners of the material; also dealing with multilingual collections. Culturally-specific offensive comments. Definition and expectations of Europeana moderation. Resources need if Europeana does the moderation.
Incentives for moderation – improving data, idealism, helping with translations – people like to help translate.

Johan's report – rewards are important – place users in social charts or give them a feeling of contributing to larger thing; tap into existing community; translate physical world into digital analogue.
Institutional policy – need a clear strategy for e.g. how to integrate the knowledge into the catalogue. Provide training for staff on working with users and online tools. There's value in employing community managers to give people feedback when they leave content.
Using Amazon's Mechanical Turk for annotations…
Doing the projects isn't only of benefit in enriching metadata but also for giving insight into users – discover audiences with particular interests.

Costis commenting – if Europeana only has thumbnails and metadata, is it a missed opportunity to get UGC on more detailed content?

Is Europeana highbrow compared to other platforms like Flickr, FB, so would people be afraid to contribute? [probably – there must be design patterns for encouraging participation from audiences on museum sites, but we're still figuring out what they are]
Business model for crowdsourcing – producing multilingual resources is perfect case for Europeana.

Open to the floor for questions… Importance of local communities, getting out there, using libraries to train people. Local newspapers, connecting to existing communities.