Powered by Trove: Opportunities and Challenges for Queensland Libraries

Author: 
Dr Marie-Louise Ayres
Publication date: 
Tuesday, 27 November, 2012
Abstract: 

This paper was delivered by Debbie Campbell, on behalf of Marie-Louise Ayres, at the State Library of Queensland on 27 November 2012. A video of the presentation is available on Youtube.

Thank you for inviting me to speak to Queensland librarians about how Trove can benefit Queenslanders and Queensland libraries. I will be covering a number of areas in this paper. I will not be talking at all about how Trove works or what it can deliver to your users, as I am starting with the assumption that you are all familiar with Trove and have already seen how your own patrons engage with the service.

I will touch very briefly on the background to Trove’s development, before delving deeper to provide you with some insights into Trove’s current content and user base, and what we know now about user behaviour and engagement. I will provide some information on what the Queensland picture looks like and what it could be.

I will then turn to the new opportunities Trove as disseminator – rather than discovery service – offers, before turning to summarise Trove’s Strategic Directions for the next three years, and how we see Trove fitting into the changing cultural and information landscape. Last, I hope to leave you with some challenges that would mean that together we can build Trove into an even better service by 2015.

The National Library has a long history of developing services to help users discover Australian collections held in cultural and research institutions across the country. Most of these services – dating back to the 1990s – were focused around specific formats, while some were subject based.

Each of these services relied on leadership from the National Library, commitment to supporting research, willingness to invest in the service of the Australian community, and national willingness to collaborate. At the time, each of the services was highly innovative, and some – like Picture Australia – were copied around Australia and the world. They were very popular with our users, and became indispensable parts of Australia’s research infrastructure.

For many reasons, we started to think about a new framework in late 2006 and worked on this idea – then called the Single Business Discovery Service – throughout 2007. The Library decided to fund the project in the 2008-2009 financial year, and in October 2008 the project officially commenced, with a dedicated project team in place. We launched Trove stage 1 in October 2009, continued development on stages 2 and 3 in 2010, and completed stage 4 – the final formal project stage in May 2011. In June 2012 we release Trove 5.0, with development continuing at a slower pace than during formal project stages.

Three years after Stage 1 release, Trove is big, in terms both of usage and content. By October 2012, we hit a record new daily high of 64,204 unique visits to the website. This is double the number of visitors in April 2011, and growth does not look like flattening any time soon. In the first quarter of this year, there were nearly 53 million Trove page views. Trove dominates access to the Library’s website, accounting for 65% of page views and slightly more of our bandwidth. Trove users are rapidly switching to mobile devices, with more than 10% of all use now on these devices, twice the percentage we saw a year ago. We have to take this change into account in our development planning.

Trove’s user engagement has been phenomenal and far in excess of what we could have expected. Once users arrive at Trove they have plenty to do, and they are doing more of it. Every month, we see growth of between 2 and 4% in each of our user engagement activities. While newspaper correction dominates – we recently costed volunteer text correction at our lowest pay level and arrived at an estimated value of around $14 million – creation of Trove lists is growing fastest of all. This continuing growth tells us that our engagement features are not ‘flashes in the pan’ that excite people for a short time and are then forgotten.

Trove is very large, with more than 315 million resources at the end of October 2012. Size alone is not all that counts; the shape of Trove’s content and use of that content are also important considerations.

At the end of the September 2012, Trove content by work type was shaped thus:

Pie chart of Trove content by work type

All statistics and graphs require explanation. In this case, what is being counted is actually a bibliographic record, so one metadata record = 1 [1] In the case of newspapers, the count is the number of articles (rather than the number of titles, or the number of pages). However, in the case of websites, it is pages that get counted. Journal articles are the most populous content type, followed by Australian newspapers and archived websites. Books follow a long way distant, and other content types are represented by very small figures.

A different view of Trove’s shape is revealed by looking at what percentage of each work type is available online (shown here in purple), and further, what percentage of each work type is Australian content AND is online (in blue). All content in the Newspapers and Archived websites zones is both Australian and available online (although some archived websites are restricted for a period of time); these zones have therefore been excluded from this graph.

Percentage of each work type is available online (shown here in purple), and further, what percentage of each work type is Australian content AND is online (in blue).

Trove promoters sometimes believe that all Trove content is available online, and that we have somehow leapfrogged all sorts of copyright and resourcing hurdles to make it so.

The reality is that roughly one-third of all Trove content is freely available online, another third is licensed content available only to patrons of subscribing libraries, and another third is metadata records describing resources which are not online and which need additional steps for access.

It is worth reflecting on what percentage of all (non digitised newspapers, non archived websites) works described in Trove are available online, and what percentage of those are Australian content.

Again, some caveats have to be considered around this view of Trove’s content. The diaries, letters, and archives zone includes material that a discriminating human being may believe belongs elsewhere. Many unpublished works such as articles turn up in this zone because when deciding which zone content belongs to, Trove has to make a best guess on the metadata presented. The percentage of diaries, letters and archives is probably skewed by records coming from OCLC’s OAISTER, which aggregates content from a large number of research institutions. The real proportion of Australian manuscripts collections actually available online is miniscule, and links to online content are to a finding aid describing the collection, rather than the actual items themselves.  

The majority of freely available online content – with the exception of newspapers, websites and pictures – is not held or made available by Australian collecting institutions. It has been contributed by many international libraries, museums, galleries, archives and universities to OAISTER from which we’ve harvested approximately 9 million records. While OAISTER can easily be searched online in its own right, Trove has integrated this high quality content source to make it more easily accessible to Australians.

A surprising amount of online Australian content discoverable through Trove is not being delivered by Australian organisations. Almost any search will reveal that many pictures, maps and journal articles are Australian in nature but contributed by others, either via OAISTER or Gale databases [2]. This reveals the extent to which Australian related materials have made their way into other collections around the world. We hope to include an Australia subset of Europeana in the next year, further increasing the amount of high quality, rare and even unique Australian online content that we can make more easily discoverable.

Caveats aside, 45% of all pictures in Trove are online and Australian, 9% of maps, 3.5% of manuscripts (mostly likely finding aids), nearly 2% of journals, articles and datasets, 1% of music, sound and video, and one-tenth of 1% of books. So there are some things to really celebrate here, and some room for thought in terms of where we might put our collective digitisation effort to boost the proportions of content that are both Australian and online.

Another way to think about Trove is to compare the shape of Trove’s content by type against the shape of use of that content. The graph below of Trove use by content type is based on the number of searches in each of Trove’s zones, which we believe is the best proxy for actual use. The graph excludes users who choose to search from the Trove home page (the ‘all’ view) which accounts for just over 10% of all searches.

Trove – work count vs usage – October 2012

There are very significant disparities between the number of items available for a particular work type, and the number of searches of that work type [3]. Use of the Pictures zone outweighs the available content by more than 5 to 1.  Music, sound and video, and books are used slightly less than might be expected, and maps around one-third of what might be expected, perhaps because of the low percentage that is available online. The archived websites zone is used at a fraction of its content (but this is largely an artefact of measuring content by pages, and of course, archived websites are collected specifically for the future). On the other hand, the diaries, letters and archives zone is used twice as much as we might expect, the percentage of searches in the people zone is in line with the available content, and interestingly the lists zone is used much more than we would expect based on its content.

Turning to the most populous content types - newspapers and articles – the disparity between the volume of content and the number of searches is quite stark. Use of the newspapers zone is 3.5 times what we might expect, reflecting the huge popularity of Trove’s digitised Australian newspapers. Even if we include use of the home page search, newspapers accounted for 76% of all Trove activity in October 2012, and this figure remains steady from month to month. This means that the Australian libraries who worked together for years on ANPLAN and who are now cooperating to digitise Australian newspapers through the Library’s Newspapers Contribution Model, have definitely delivered content which is of very high value to the Australian community.

In contrast, use of the Journals, articles and datasets zone is vanishingly small compared to its heft in the database. The content of this zone includes descriptions of journals, and freely available articles such as those available through OAISTER. However, 125 of the 141 million items discoverable through this zone are licensed articles contributed by Gale and Informit, with more than 124 million of those 125 million from Gale.  These resources are only available to patrons of subscribing libraries, and were selected because they were included in the NSLA core set [4] off eResources (some additional Gale databases not in the core set were also included [5]).

The Library completed a review of the Trove articles service in August. We found that the infrastructure we built is effective for connecting users with content which they are entitled to use, that use of the content via Trove is healthy compared to what we know about use via some individual libraries, and that use of the Trove articles service has doubled over the last year.

However, we were very surprised to discover that the service is not reaching its target audience. We built this part of Trove primarily to benefit patrons of NSLA and public libraries. However, we found that the majority of service use was by patrons of university libraries, who are generally well served by their large research libraries and sophisticated discovery layers. We also found that adding and maintaining additional databases in Trove – which has no revenue stream and is entirely funded from the Library’s recurrent budget – is beyond the Library’s means. The Library has therefore decided that we will not expand the service to include additional vendors or databases. We are extremely grateful to both Cengage Gale and RMIT Publishing for their willingness to work with us in these very untested waters, and to continue to work with us in the future.

Trove has three major sources of content. The first is the Australian National Bibliographic Database (ANBD). Every record that is in Libraries Australia – with the exception of Serials Solutions records – is pushed through to Trove. So for libraries, the most important Trove contribution pathway is via the ANBD. Libraries Australia has contributed 23 million records to Trove and this number increases every time an Australian member library adds a bibliographic record to the ANBD. Trove’s very high usage mean that exposure of records through the service represents a significant additional return on investment for cataloguing effort.

The second is the Newspapers Digitisation Program. We had 7.5 million pages delivered through Trove at the end of October 2012, and that number will be be boosted significantly over the next 3 years. We expect to deliver at least 3 million more pages by 2015, thanks largely to a very large State Library of New South Wales digitisation project funded in the last New South Wales budget, being managed through the National Library’s contribution model. The National Library will shortly commence mass digitisation of books and journals, and it is possible that a similar model will see collaborative digitisation and delivery of more Australian published content in the future.

The third source of nourishment is the work we do to harvest records describing digital collections into Trove. Aside from the ANBD, Pandora and newspapers, 109 contributors add their content directly to Trove, with those contributors indicated below, in blue by state:

 Trove – current vs potential contributors

Universities account for a high proportion of these 109, as all universities contribute their research outputs to Trove. Libraries account for another significant component, with a fairly modest sprinkling of galleries, museums, archives and other cultural organisations also contributing. Queensland has more contributors than might be expected based on its population, thanks largely to the State Library of Queensland’s successful Picture Queensland project. We also have a number of international contributors, including OCLC’s OAISTER service, a subset of the Hathi Trust digital imprint, Open Library, and Cengage Gale.

The red column indicates potential contributors; those who have contacted the National Library seeking to be included, or those the Library has identified as having high quality collections in scope for the service. There are many more potential than actual contributors (and of course there will be other potential contributors unknown to the Library), with the great majority being in NSW.

In the 2011-2012 financial year, the collections of seven major contributors were added to Trove. This year, we will add many more by working with groups of potential contributors who either have a similar business need (as is the case for universities, many of whom have a funding imperative to contribute researcher data this financial year), or who use a similar platform (for example, we are currently working with a number of small museums who use the web-based eHive collection management system).

The following Queensland organisations currently contribute directly to Trove.  The State Library of Queensland and the universities also contribute via Libraries Australia.

  •        Banana Shire Library and Information Service

  •        Barcaldine Library

  •        Bundaberg Regional Library Service

  •        Burdekin Library

  •        Cairns Libraries

  •        Central Queensland University Library

  •        City Libraries Townsville

  •        Fraser Coast Regional Libraries

  •        Griffith University

  •        Gympie Regional Libraries

  •        Hinchinbrook Shire Library

  •        Ipswich Library and Information Service

  •        James Cook University

  •        Logan City Council Libraries

  •        Monument Australia

  •        Moreton Bay Region Libraries

  •        Noosa Council Library Services

  •        Performing Arts Centre

  •        Primary Industries and Fisheries eResearch

  •        Queensland Museum

  •        Redcliffe City Library

  •        Scenic Rim Regional Council Library Service

  •        State Library of Queensland

  •        Sunshine Coast Libraries

  •        Tablelands Regional Library Service

  •        Queensland University of Technology

  •        University of the Sunshine Coast

  •        Western Downs Libraries

Queensland universities contribute records from their research repositories and – increasingly – records about their researchers through the Trove party infrastructure, including the Trove Identity Manager (TIM). Primary Industries is also contributing records on its research outputs.

Most other Queensland contributors are public library services, with only a few from other sectors. In Queensland – as is the case across Trove – it is primarily libraries which have seen the opportunities Trove and its predecessors offer, and have the institutional will and technical know-how to contribute.

It is important to note that our list of potential contributors includes everything from ‘we think it’s a good idea’ or ‘they have contacted us’ to ‘they are just about ready to contribute’. The list of potential Queensland contributors provides a good ‘snapshot’ of the various states along the contribution pathway. Because of my background with archives, I see this list of potentials as very much like a list of potential Manuscripts acquisitions – in some cases, the organisation has contacted us, in others we know they have content of interest so will make the running when we can, and there can be long gaps between first, second, third etc. contact. People working in organisations move on, or their priorities change, or they just forget. And sometimes both we and they do a lot of work without getting to the ‘payoff’, when their content is visible through Trove.

This is a view of current Queensland contributors, mapped to their locations.

 View of current Queensland contributors, mapped to their locations

Even if we added our seven known potential Queensland contributors, this picture would not change much. Unsurprisingly, they are clustered around the south-east corner, and there is nothing north of Cairns. It is possible that some Queensland public libraries are contributing content such as local image collections via Libraries Australia. But we suspect there are plenty of community collections in libraries further from that south-east cluster, which would be in scope for Trove.

We simply cannot travel around Queensland, but one thing those of you who do visit rural and regional public library locations could do, is to ask whether the library is a member of Libraries Australia. If so, does the library have a local digital collection and do they contribute it to Libraries Australia or Trove? Perhaps the library is not a Libraries Australia member but has a local digital collection that they do not know how to or have not yet thought about contributing to Trove? If we found there were many libraries which have collections that would be useful to include in Trove, how could we go about communicating with them about what they need to do to contribute? And if they have a lack of technical know-how, what could you or we do about that?

Most non ANBD and non newspapers contributions are managed and undertaken by the Trove support team under Debbie Campbell’s direction, with some – especially really big datasets such as OAISTER and Gale – requiring specialist IT resources. We acquire content using a number of mechanisms, with the Trove harvester an essential piece of infrastructure to support smooth addition and updating of contributors’ content.

One of the Trove team’s realities is like the one faced by special collections colleagues – that while we would like to much more pro-actively ‘shape’ the collection, much of what we do is reactive. If we are approached by an organisation that seems ready, willing and able to give us their content, we will work with them, and perhaps put on hold an organisation with high value content we really want, but where we detect a lack of organisational will or capacity.

The reality is that we often have to make decisions about adding easier to get but possibly lower value content or harder to get but very significant content. We do not forget the difficult ones – and we do get wins from time to time – but this is always a balancing act for the Trove team.

I would now like to turn to the user perspective.

At the moment, we do not know enough about Trove users, how they view their experiences, and what they need or want. One thing we do know is that 35-40% of users come from overseas. Those visits (which are on average much shorter than visits by Australians, with a higher bounce rate) no doubt promote Australian culture which is an excellent outcome. However, our focus at the Library is on providing services to Australians. 

We do gather some information about Trove users who send us enquiries, compliments or complaints. We know what the Twitterverse says and are often lucky enough to stumble across what users have written or said about Trove, or about how they are using or intend to use it.

We know how many users visit every day, what kinds of devices and browsers they are using, and we know some things but not enough about what they do when they are visiting. The Library’s Paul Hagon recently analysed use of the Library’s website, catalogue and Trove. Trove users ‘stick’ for an average of 8 minutes 45 seconds per visit, compared to 3 minutes 11 for visits to the Library’s website and 1 minute 19 for visits to the Library’s catalogue. Australian users spend even more time, averaging 12 minutes 14 seconds on the site. Trove is what technologists call a very ‘sticky’ site.

We know that some Trove users are absolutely committed to correcting newspaper text, and that research is happening right now on patterns of correction; Paul Hagon will give a paper on this at the ALIA Online conference in Brisbane in February 2013. Research is also underway on some of the motivations behind this kind of community engagement [6].

We know that Trove users align quite closely with actual population distribution in our states and territories, although Queensland use is lower than expected based on population.

 Graph of Australian Trove usage vs actual population

From what we can tell from Google Analytics, use by non-metropolitan users is below – but not alarmingly below – population distribution. Non-metropolitan use seems to be particularly healthy in Queensland, mirroring our enquiries from Trove users; just over 50% of QLD enquirers who told us their postcodes were from outside the metropolitan and Gold Coast regions.

But these are really just glimmerings of understanding, and there so much we do not know. We will be undertaking a large user evaluation this financial year to help us discover who our users are, what they are seeking, whether they find it, where else they might find it and we hope many other things to help us to understand our Trove user context and to build our services and content accordingly. We will report on the outcomes in the 2013-2014 financial year.

I have focused so far in this paper on what is in Trove, how it gets there, what the Trove content picture looks like, and what we know about Trove users. But with the delivery of the Trove API in April 2012, Trove moved beyond being an aggregator and discovery service, to being an open data store available for re-use outside Trove. Trove makes almost all records and all digital content available for re-use through the API, including newspaper text, user corrections, tags and notes. Non-commercial users do not need to seek our permission to use the API for their own purposes – although we do like to hear about it. Commercial users need to contact the Library to discuss their proposals prior to being issued with an API key.

Dr Tim Sherratt, digital historian and developer of digital history tools, QueryPic is a good example of a Trove API user. Tim built this tool to map occurrences of words and phrases in newspapers, to build up his own picture of the ‘mood’ of Australia at various time, and for many other kinds of historical enquiries (when, for example, did the front page of newspapers become the place where the main news of the day went?). Although Tim started out scraping Trove content, his job has been made much easier with the advent of the Trove API. This QueryPic graph shows occurrences of the word ‘trove’ in all Trove newspapers, by decade.

 This QueryPic graph shows occurrences of the word ‘trove’ in all Trove newspapers, by decade.

This view is as a percentage of all available content, rather than actual numbers. The high point for the word ‘trove’ to 1955 was in 1936, when it appeared 1009 times. It seems that a lot of ‘treasures troves’ were dug up from under old houses, old pubs, and town halls that year. I like to think that we’ll have brought the word back into more common use over the last few years.

Tim’s tool finds occurrences of words and then presents them nicely for consideration. Other researchers have been experimenting with Trove’s very large text base to develop programs that can ‘learn’ certain kinds of associations that go beyond words and into concepts. In three years’ time, I believe will see a number of these ‘power’ uses starting to bear fruit. 

Other researchers have used Trove’s API to experiment with alternative ways of presenting Trove images. Dr Mitchell Whitelaw, an Associate Professor at the University of Canberra is working with a team of researchers on their quest to create richer visual interfaces to image based content. With funding from the State Library of New South Wales, Mitchell used the Trove API to gather 7000 images from the Manly Local Studies collection, and created a visual interface which does not rely on search.

Most of our libraries –– including the National Library –– look something like this to our users:

Screen shot of Manly Local Studies Image Library search interface

It is not surprising that imaginative designers ask themselves whether there is a better way, nor is it hard to imagine our collective users enthusiastically taking to browse interfaces like the the Manly Local Studies Image Library, which uses the API to extract a curate content set:

Screen shot of Manly Local Studies Image Library search results interface

Another Whitelaw prototype, Trove Mosaic, was released in November 2012. The prototype takes the first 500 search hits it finds by using the API and then builds rotating mosaics to display that content.

Screen shot of the Trove Mosaic search results interface

These examples pull Trove content out to external services, or pull out a dataset to manipulate using analytical tools. But Trove content can also be included in existing services. We have long anticipated that an Australian library might want to include content of interest to local patrons but held by other collecting organisations in their own websites. This has not yet happened yet but we hope that it will. Given that the CEOs of the National, State and Territory libraries have agreed to offer a prize for the 2013 GovHack, we may well see Trove content turning up in an existing service as part of the June long weekend event.

One example – outside the Library website arena – was developed by the Library’s Paul Hagon over the course of a long weekend, after the VALA conference in February. Paul built a WordPress Plug-in which uses the Trove API to find and display text from specific Trove newspaper articles. The text is displayed in WordPress blogs with full citation – and if the text is corrected in Trove, it is automatically updated in the WordPress display.

It is important to note that there are simpler ways of being powered by Trove. For example, Queensland Government Agency Libraries (QGAL) have created a Trove user identity for QGAL, adding all QGAL libraries to the ‘My libraries’ list. When they search Trove and limit their search to ‘in my libraries’, they can view holdings just from the QGAL group. This is a really simple way to create an effective union catalogue search.

In this paper, I have tried to give you a clearer picture of Trove’s shape, its potential shape and some ideas for how it might generate specialist new services, or augment existing services, or even shape how we can deliver Trove in the future. Trove’s Strategic Directions for 2012-2014 reflect those ideas for the future. Trove’s Strategic Directions were developed by the Trove Reference Group which includes wide representation from across the Library. They arise directly from and are closely aligned to the National Library’s Strategic Directions which has four key goals:

  1.      Collect and preserve documentary heritage

  2.      Make our collections and services accessible to all Australians

  3.      Deliver national leadership; and

  4.      Achieve organisational excellence

Trove’s Strategic Directions are most closely aligned with the 2nd and 3rd of the Library’s directions. They reflect the Library’s understanding of the current and likely 3 year context in the information and cultural spheres. We expect many more Australians to have access to fast broadband by 2015. We expect that a large proportion or even a majority will be using mobile devices to access content. We expect users to be able to choose from an ever-growing array of online content, and will therefore focus on what we, uniquely, can do, including digitising Australian content, and making freely available digital Australiana more discoverable.  

We expect a new Australian Government Cultural Policy to be launched, and that it will extend beyond the performing and creative arts and into the areas of how Australians can access, use and interact with their documentary heritage. We support the Australian Government’s focus on inclusion, especially ensuring that rural and regional Australians have access to the information and opportunities they need. We know that the school education landscape will be in a period of considerable change, both as a result of the Gonski review of education funding and through implementation of the national curriculum, supported by classroom ready digital curriculum materials. We are delighted that Minister Simon Crean announced welcome additions to our base funding for each of the next several years in the May 2012 budget, which we will put to good work in improving our digitisation and delivery output. We do not expect national licenses for eResources to come to pass in the next three years, a key factor in our decision to put less effort into including licensed article content, and more into our niche role of freely available digital Australiana.

For each of our Trove Directions we have described our aspirations and the actions we plan to take over that period. We are currently implementing our new Trove Information Architecture and will publish our Directions online when that is complete. To support building Trove’s content, we have just completed the Trove Content Inclusion Policy, which will also be published soon on the website. We are thinking about how we can provide one to many instead of one to one advice to potential contributors – in other words, how we can efficiently provide advice so that we can work with a pool of self-sufficient and ‘Trove-ready’ contributors.

In an effort to build better services, we will enhance Trove interfaces so that all parts of the service are mobile friendly, beginning with the Australian newspapers interface in 2013. In early 2013 we hope to deliver a ‘selected subject view’ covering the performing arts, replacing some of the specialist functions we offered through Music Australia and Australia Dancing. We will work with our NSLA Delivery project colleagues to offer good pathways to fast, reliable and efficient document delivery services for resources which can be discovered through Trove but have not yet been digitised. And we will support and strengthen our online engagement with the Trove community through multiple social media channels, including some new ‘how to’ videos this financial year.

Building a sustainable business model for Trove – so that we can enrich its content and invest in new service options in a scalable way and within available resources – is a challenge, and we will work hard to build capacity in the potential contributor pool to reduce overheads associated with adding new content and maintaining existing contributions. We will investigate diversifying our funding base beyond the Library’s recurrent budget and those of libraries contributing digitised content – but we are not under any illusions about the nature and scale of this challenge.

I’d like to leave my Queensland colleagues with some questions or provocations that I hope you will turn into concrete actions.

Starting at home is always the first priority! We find that librarians and especially CEOs are often far too optimistic about the extent to which their collections are exposed in Trove. So first things first - does your library contribute regularly and fulsomely to Libraries Australia and/or Trove? Do you know exactly what proportions of resources held and described by your organisation have been contributed to the national services? If you have described special collections or digitised content and you have not contributed them to Trove, why not? What is stopping you? Can you overcome organisational, technical or financial barriers that are preventing your collections from appearing alongside content from across the nation?

Now that you know more about what Trove content is available online and is Australian, and what kinds of content currently attracts most Trove use, how will you use this information to drive your decision making about building your born-digital collections, and to digitise your physical resources? Now that you know that Queensland and non-metropolitan users are every bit as important to us as they are to you – and that Trove use by Queenslanders is below what we might expect based on population – what unique and regional collections do you think should be the focus of your digital effort? We need every community in Australia to feel that it matters, and to be able to both contribute to and see its own story in Trove.

The disproportionate use of pictures and manuscripts suggests there is a hunger for this material, but the State Library of Queensland recently found that on an item by item basis, use of digitised publications significantly outweighed use of other content types. If we can increase the number of Australian publications that are actually online, will this drive increases in use of this kind of content?

Beyond your own organisations, you all have local networks, most of which are far beyond our reach in Canberra. Do you know of content – especially freely available digital Australiana – that would be perfect for Trove but is sitting, less discoverable than it should be – in individual databases or websites?

Could you spread the Trove word, to convince those with suitable content – especially digital content – that participating in a national service will benefit them and their users, rather than posing a threat to their individual branding (something we often hear from organisations other than libraries)? Can you convince them to think local but act global? In your own networks, how could you incorporate the Trove message in your regular communication mechanisms? How could we help you to be clear about that message and to work with you to start conversations with potential contributors?

Encouraging contribution is all very well, but goodwill does not get us over the line. Many organisations with good digital collections lack the capacity to expose them to any aggregator, including Trove. They have limited staff, or are run by volunteers, and have no technical capacity. They are unable to set up a simple HCL mechanism, or to set up OAI-PMH repositories to allow us to harvest their data regularly. Is there anything you can do about this? Do you have the know-how in your organisation and could spread the love? Do you know of any volunteer or low cost programs that match IT specialists to worthy organisations, especially those in more remote areas? Could you advocate for such a thing – a process where organisations could source, at no or low cost – advice to help them get to the point where we can actually start working with them?

75% of all Trove users arrive via Google, so it is clear that our most effective marketing strategy is to expose Trove content to search engines. But we know that many Australians who would benefit from Trove do not know about it, or how it might help them with their enquiries. We do not have an advertising budget and rely on our community to advance knowledge of Trove and what it offers. Many of you will already have strategies in place to promote Trove to your local patrons, whether you are engaging with them onsite or offsite. If you do, do your strategies need a rethink? With so much content now available through a national service rather than our individual websites, do you need to rethink your messaging? Do you need to see your own website – and all its local content important to your users – alongside Trove?

How can you provide us with feedback about what your local experiences tell you Trove users want and need? I mentioned earlier that we will be undertaking a large scale user evaluation this financial year, and we will certainly be sharing the results with the library community. We definitely do not want you to communicate every Trove like and dislike from your users – we would be swamped! But what do you think would be the best mechanisms for you to provide feedback to us and to share information about how your users interact with Trove?

Having now rather importunately asked you to do a whole lot of things for Trove, I now ask about how you could use Trove to achieve the needs of your users.

The developers of the Trove-powered tools I have mentioned in this paper agree: the key ingredient to exploiting the Trove API is not programming skill – it is imagination and inventiveness, and the ability to move from ‘what if’ to actually having a go. It is the same kind of imagination we see at work in the burgeoning App world, including identifying needs we did not even know we had, and then engaging us with the opportunities they open up.

The first step in thinking about how your organisations could be powered by Trove may be to brainstorm, to be as wild as possible in thinking about how you could put Trove to work for your users. Play around, draw up some wireframes, and get your bright young and not-so-young things to think about what will deliver most value for your users.

At some point you will to need to decide whether you want to integrate Trove content into an existing service, or build a whole new web-based service. For example, would the State Library of Queensland want to dynamically include Trove content in its Queensland Convicts information, so that any new Trove content – including user tags and comments – would automatically update in the SLQ site?

Or is there a regional library service that would like to present a local ‘view’ of photographs, newspapers, even oral histories that relate just to that area. For stand-alone services, you will need to think about the lifespan and sustainability of your new service. We will keep making the content available – but you will need to think about whether your new service is a prototype, or has a 3 year or a 5 year lifespan.

You will then need to think about how you are going to go about it, how to utilise existing skills, borrow them, or ‘buy them in’ (which is what happened with the Manly Image Library example. Of course, if you are thinking about anything which will be large-scale and ongoing in nature, you will need to contact us early so we can establish whether what you are proposing will jeopardise Trove’s performance, or will fall outside uses that the Library feels are acceptable.

So, our current Strategic Directions runs from 2012-2015. It’s still hard to know exactly how many of our specific actions we will achieve, and how well we will have met the Directions by then.

But if I ask myself how I will know whether this talk to you today has been successful, some of the things I would like to see are these:

  • Absolutely every State Library of Queensland holding, for every format, included and current in Trove

  • More digital Queenslandiana – newspapers, pictures, websites, ePublications, music, moving image

  • Absolutely no digitisation programs – especially for newspapers – which are incompatible for inclusion in Trove

  • More local studies collections held by public libraries – especially outside the biggest population centres digitised and available in Trove. Should we put a number on it? Another 10 in 3 years time? Or think about what that map of Queensland should look like in 3 years time? Would we see more satisfying little flags out beyond the high density belt?

  • I would like to see so much unique and wonderful Queensland content that Trove becomes irresistible for Queenslanders, and use grows to levels that would keep me and my IT colleagues awake at night, wondering how we can augment our infrastructure to match demand.  

  • I would love to see use patterns that match Queensland’s population distribution, that mean that people living in rural and remote areas are just as likely to use Trove as their city cousins. If I was travelling through country towns and remote areas wearing my Trove T-shirt, I would love to be stopped by people who want to tell me how Trove has changed their understanding of the world, and of their world. 

  • And last, I would like to see some fully established ‘powered by Trove’ sites, where you can build new services or augment your own by including Queensland related content that we have aggregated from others.

Which I think means perhaps I should talk to you again in 3 years time!

Footnotes

[1] This method of measuring content has obvious drawbacks. A large archive running to several hundred boxes may be represented by a single metadata record, as may a single journal article, a single ‘page’ in a website etc. However, consistently using the same method gives us a good picture of changes in Trove’s content shape over time.

[2] The content of five Gale databases is available via Trove to registered users of subscribing libraries.

[3] Clearly the method of counting content and counting usage influences the shape of this graph, and that alternative methods would yield different results. For example, if archived websites were counted by title archived, rather than individual pages, the disparity between content and use would be far lower. However, if digitised newspapers were counted by pages rather than article, the disparity between content and use would be much higher. Overall, we believe the approach taken is as valid as any other in determining which kinds of content are of most interest to Trove users, and what relation that level of interest has with the amount of content available.

[4] The National and State Libraries Australia eResources Consortium aims to make a ‘core set’ of eResources available to all registered patrons of the National Library, state and territory libraries.

[5] At October 2012, content from the following databases was available via Trove to registered patrons of any library subscribing to these products: Gale Academic OneFile, Literature Resource Center, Health and Wellness Resource Center, General OneFile, Expanded Academic ASAP; Informit Australian Public Affairs Full Text (APAFT), Business Collection, Engineering Collection, Families and Society Collection, Health Collection, Humanities & Social Sciences Collection, Meanjin, Media International Australia, New Zealand Collection, Indigenous Collection and Literature and Culture Collection.

[6] Alam, Sultana Lubna, and Campbell, John, ‘Crowd sourcing motivations in a not-for-profit GLAM context : the Australian newspapers digitisation program’, a paper presented at the 23rd Australasian Conference on Information Systems held at Deakin University 3-5 December 2012. The paper is available from the Deakin University research repository, and was discoverable via Trove within days of its delivery, demonstrating Trove’s power in disseminating new research.