The Five Stars of Open Access (aka Linked Documents)

Yesterday I was having a discussion about Scholarly Communications, Open Access, Web 2 and the Semantic Web with some colleagues in our newly formed "Web and Internet Science Research Group" at Southampton. As we were comparing and contrasting more than a decade's experience of open access/open data/OER/Open Government Data, we made the following observation: reuse is the enemy of access.

There have been efforts to replace PDF with HTML as a scholarly format to make data mining more easy, and movements to establish highly structured Learning Objects rich in pedagogic metadata to facilitate interoperability of e-learning material. (I have been involved in both of these!) But both have been ignored by the community - they are too hard, they fly in the face of current practice, they involve users learning new skills or making more effort. Some would argue that similar comments could be made about preservation and open access, or even just repositories and open access.

Although "reuse is the enemy of access" is quite a bold statement it's really just a reformulation of the old saw "the best is the enemy of the good". Attempts to do something with the material we have available are always more complex than just looking at the material we have available. Adding services, however valuable and desirable, are more problematic than "just making material available". In the repository community we've worked hard to help users get something for nothing (or something for as little effort as possible), and I'm proud that people recognise that philosophy in EPrints. But it's still a tension - you have to present Open Access as a bandwagon that's easy to climb on!

So I'm particularly impressed with Tim Berners-Lee's Five Stars of Linked Data as a means of declaring an easy onramp to the world of Linked Data, while at the same time setting out a clear means of evaluating and improving contributions and the processes required to support them. It allows the community to have their cake and eat it; to claim maximum participation (a bigger community is a more successful community) and appropriate differentiation (better value is a better agenda).

I think this approach would have served the Open Access communities (OA/OER/Open Data) very well (why didn't we think of it?) But it could yet do so, and so in the spirit of reuse I offer some early thoughts on the Five Stars of Open Access.

★ Available on the web (whatever format), but with an open licence

★★ Available as machine-readable editable data (e.g. Word instead of PDF page description)

★★★ as above plus non-proprietary format (e.g. HTML5 instead of Word)

★★★★ All the above plus, use open standards from W3C (RDF and microformats) to identify things, so that people can understand your stuff

★★★★★ All the above, plus: link your data to other people’s data to provide context i.e. link citations to DOIs and other entities to appropriate URIs (e.g. project names, author names, research groups, funders etc).

These are directly taken from Tim's document, with some subtle variations, and are intended for discussion. For a start, it shows that we haven't even got very far into 1-star territory, as we mainly fudge the licensing issue. (This comes from the fact that unlike data, our documents are often re-owned by third parties.) Pressing on, the second star is available for editable source documents rather than page images and this is also a minority activity. In our repository, there are 7271 PDFs vs 820 Office/HTML/XML documents. So a long way to go there. The third star seems even more remote (376 documents). And as for the fourth star's embedded metadata?

But the fifth star: this seems to be so valuable. If we could just get there - properly linked documents, no chasing down references, the ability to easily generate citation databases, easy lookup of the social network of authors. Sigh. What's not to like? And you can even add 5* facilities to PDF, so perhaps we will find some short cuts!

If we develop these five stars, it will help us to function as positive Open Access evangelists, while also promoting the future benefits that we would like to work towards. No mixed messages. No confusion.

Open Access - Who Calls the Shots Now?

Three years ago on this blog (doesn't time fly!) I contrasted the efforts that librarians and academics could make in furthering Open Access. My argument (such as it was) focused on the relationships between the two communities, noting that when it came to research, librarians could only advise and assist but that academics could lead and command. Or at least in theory! In particular I backed the idea that change would come from senior managers in the academic world and from research funders. In the intervening time we have indeed seen a big increase in OA leadership in the form of mandates being adopted, but I wonder if the pace of change is not about to put even researchers in the back seat.

The Web was developed at CERN, in Switzerland, and took over the world in more than a geographic sense. It emerged from its home in a highly-funded, very collaborative, international research laboratory and carried the culture and design assumptions of its birthplace (open information exchange, minimal concern over intellectual property control, no requirements for individuals to monetize knowledge production) and stamped them on the rest of society, regardless of society's estimation of its own needs (for more, see the presentation The Information Big Bang & Its Fundamental Constants). One manifestation of the clash between the Web and "how society has historically operated" was the Budapest Open Access Initiative some ten years after the initial development of the Web.

The Web's culture of open information exchange has more recently had a very visible effect in the area of Open Government Data. A simple re-statement of the objectives of the Semantic Web as The Five Stars of Linked Data has powered a tremendous focus of activity in national and local government when allied with political agendas of Transparency and Accountability. Portals like data.gov.uk and data.gov provide access to "the raw data driving government forward" which can be used to "help society, or investigate how effective policy changes have been over time". In the UK, the Treasury's COINS database of public spending is one of 5,600 public datasets that have been made available as part of the initiative. In the US, the Open Government Directive requires each department to publish high value data sets and states that "it is important that policies evolve to realize the potential of technology for open government." Both US and UK government see the opening up of public data as the driver for political improvement, innovation and economic growth, with the Public Data Corporation as the focus of British development of an entire social and economic Open Data ecosystem.

Having watched Open Access lobbyists engage in political processes in the UK and US (with a handful of Senators, Congressmen and MPs sometimes for OA and sometimes against) it is rather a shock to see the President and the Prime Minister suddenly mandating a completely revolutionary set of national policies based on the technological affordances of the Web, and in the teeth of plenty of advisors' entrenched opposition. And rather a shock to realise that offices even more elevated than a vice chancellor are enthusiastically joining the world of open resources and open policies.

But data and publications are different things, and publications are privately owned by private publishing companies rather than stockpiled by the government. However, the decade of Open Access debate has shown that progress in OA (and OER and open data) is impeded more by individual and institutional inertia than corporate opposition. When the highest offices of government are confidently pushing forward a programme of open participation, will academics have the luxury of treading water?

How will our governments sudden enthusiasm for open data affect Open Access? Perhaps not at all. Perhaps Universities are too insulated from the administrative whims and shocks of Washington and Whitehall to be affected. (How many researchers have even heard of data.gov?) Even so, governments will indirectly cause a shakeup in the administration of public research funding, and the infrastructure needed for universities to adequately respond to the requirements of open funders will cause them to become more open themselves.

The public climate that informs the private OA debates and decisions in University boardrooms will change; pro-OA researchers and librarians will no longer be arguing from such a defensive position, not appearing as idealistic hippies. Even in the absence of direct government mandates, pro-OA decisions will be easier to support and less contentious to implement. The values of the research communities will change as public values and expectations change - when even governments become more accountable through open data, research communities that insist that their data and their research is their private property, for the sole benefit of the furtherance of their own careers, will soon appear old-fashioned and untenable.

So watch this space. It may be that Cameron and Obama will indirectly achieve what Harnad and Suber have been toiling for. I wonder what I'll have to say in another three years' time?

Rehabilitating The Third Star of Linked Data

The mantra of open data is: put your data on the web / with an open license / in a structured, reusable format / that is open / using open identifiers / that are linked with other data.

The third step/star in this process is commonly explained as using CSV rather than Excel, (because the former is an open format, but the latter is a closed proprietary standard). You'll see this position stated at Linked Data Design at the W3C and sites all around the world are copying it.

We really need to think a bit harder about this: Excel's native format is an open standard, and although an XML encoding of a the complete semantics of a spreadsheet is hardly a straightforward thing to deal with, it is simple enough to extract data from. In particular, I don't see that it is significantly more difficult than dealing with CSV!

Once you've unzipped the Office Open XML data, you can iterate around the contents of the spreadsheet, or extract individual cells with ease. And without any .NET coding or impenetrable Microsoft APIs. Here's a simple example that lists the addresses and contents of all the cells in a spreadsheet.

<xsl:template match='/'>
<xsl:for-each select="/worksheet/sheetData/row/c">
<xsl:value-of select="@r"/> = <xsl:value-of select="v"/>
</xsl:for-each>
</xsl:template>

Of course it's simplified: i've missed off the namespaces, and strings are actually stored in a lookaside table and there are multiple sheets in a single document, but even so I'd rather wrangle XML than wrestle with CSV quotes any day.

Response to An Open Letter on OpenAIRE

I hope that my friends and colleagues in the Open Access movement will forgive me for the following words, written in response to John Willinsky's blog post The Enlightenment 2.0: An Open Letter on OpenAIRE:

Oh punner! While you are Open,
Err on the side of caution.
O, pen airy words of knowledge
To fill us with a sparc of hope an' ne'er
Leave us Else ever 'ere.

I'll get my coat.

Visibility of OER Material: the Jorum Learning and Teaching Competition

The recent ALT-C 2010 conference saw the final six winners in the Jorum Learning and Teaching Competition present their resources, and receive their prizes, with those taking the top three places announced at the gala dinner.

Louise Egan, JISC-Repositories Email

This competition was designed to promote people sharing learning resources, and that's fantastic. Since none of the six winners' resources are actually deposited in the Jorum repository (just metadata records with a web link to the actual location), this situation provides an interesting insight into the advantages of depositing material (or a reference to material) into a repository. Since copies of all the winners material reside elsewhere on the Web, can we find out which link gets the higher ranking: Jorum or the original university? Is there any consistent pattern that emerges? In the following breakdown I'll list the Google rankings of each resource when searching for the title of the resource.

First prize: The Molecular Basis of Photosynthesis

There are actually three places to find of this work: Jorum, the creator's personal website and a Cambridge support site which contains the original version of this resource (one that hasn't been split into separate sections).

Jorum: 3,4

Author: 6

Institution site (cam.ac.uk): 1,2

Second prize: The Open Dementia E-learning Programme: Living with dementia

Jorum: 3

Institution (scie.ac.uk): 1,2

Third prize: Making the Creative process visible

The home of this material is on Vimeo, although it is also referenced by the HEAcademy who sponsored the project

Jorum: 1,2

HE Academy: 3,4

Vimeo: 6

Fourth Prize: Ayo Gorkhali

Jorum: 3,4

Professional Society Page: 1,2,7

Fifth Prize: Interpreting Skills Map

Jorum: 3,4

Professional Society Page: 1,2

Sixth Prize: Plagiarism Tutorial

There are a LOT of Plagiarism Tutorials offered by Universities all over the world!

Jorum: 66

Institution: 89

So in some cases JORUM boosts the ranking, and hence the visibility, of an item, whereas in other cases it doesn't. Can we draw any general patterns from this small sample? To be honest, I don't think so! The range of institutions is too diverse. Some of the alternative locations are highly visible, so it is not surprising that Jorum is eclipsed by their ranking (e.g. Cambridge, very newsworthy Gurkhas international organisation). Some 49% of Open Jorum's records provide links to external sources rather than holding bitstream contents directly. It would be very interesting to see the bigger picture of OER visibility by undertaking a more comprehensive survey.

More on Mendeley and Repositories

Yesterday's post Comparing Social Sharing of Bibliographic Information with Institutional Repositories created a few comments, so I thought I'd make some more observations from an outsider's point of view.

I think that Mendeley are a fascinating example of the Open Access problem. OA is about moving knowledge from researchers' private environments (their laptops, hard disks, CDs and filing cabinets) into the public space (repositories, websites, search engines). Mendeley's software spans both those environments - bibliography management for the desktop feeding researcher profiles and CVs on the Web.

As Victor Henning pointed, Mendeley are part of Cambridge's JISC DURA project, which aims to take advantage of Mendeley's position bridging the desktop/Web to try and encourage more public repository deposits. This is a very interesting proposition: maybe a users of a such a service will be more inclined to make their work Open Access? Perhaps the simple act of buying into the "Mendeley proposition" will cause them to be be more favourable to Open Access than they would otherwise have been?

From the outside it's difficult to understand the extent of Mendeley's penetration into a University. What is visible is the public profiles that Mendeley users have created. Although the Mendeley API doesn't allow searching for users, I have been able to identify 53 public profiles from the University of Cambridge through Google (and a lot of manual verification!) Incredibly, only TWO of those 53 researchers have any existing deposits in Cambridge's institutional repository.

This is potentially great news: Mendeley's software has gained takeup from users who aren't repository users. They aren't preaching to the converted, they are getting new users to work in the open, to start to make the transition from the desktop to the Web.

But the OA battle hasn't been won yet. Of those 53 profiles, 21 contain no publication information, and of the 32 list their publications, only 9 have made any of their publications open access through the Mendeley service (a total of 40 PDFs).

The social bibliographic approach that Mendeley are promoting is a promising way forward. It's offering people something that they haven't seen from the repository, but it's not a principally Open Access offering, and it's no silver bullet for providing open access. Commentators who have suggested that repositories are old-fashioned, and that everything can be solved by Web 2 solutions, are being over-optimistic. Repositories are hard work because changing researchers' working practices is hard work and I guess there's no single magic solution that's going to make that effort disappear!

Comparing Social Sharing of Bibliographic Information with Institutional Repositories

Everyone seems to have been talking about Mendeley over the past year! They have won a string of prizes, most recently the Guardian "Activate Future Technologies" workshop award for the project most likely to change the world for the better. They have achieved these accolades by providing bibliographic database software for the desktop ("like iTunes for research papers"), coupled with a social web site through which researchers can share their bibliographic collections. They have been successful to the tune of 47,5671 users and 34,852,751 documents (according to figures on their home page), with some commentators suggesting that they may soon provide access to more bibliographic data than Thomson ISI!

Now a lot of these "documents" are private material that are just stored on researcher's desktops. I am not interested per se in which software is being used to manage private bibliographic metadata. But the extent to which the "social sharing" agenda is successful is obviously crucially important to the repository community - to what extent is research being shared publicly, and in particular, to what extent are full texts of scientific papers being provided as Open Access through Mendeley's site.

To investigate these issues, I took a snapshot of some of their user profiles. Mendeley have 33678 public Computer Science profiles listed, so I took a 10% (3423) sample of those. Of that sample, 2918 or 85% have no publications listed at all, while 6% have only 1 or 2. Just 2% have 10 publications or more listed. The whole sample has a total of 2317 publications listed, with 681 providing PDFs from the Mendeley website. If this sample scales up (and the method I used does not constitute a proper random or representative sample), then the computer science part of Mendeley would have about 23,000 publications listed, with just shy of 7000 full texts.

By contrast, our departmental repository (eprints.ecs.soton.ac.uk) has about 15,500 publications listed, of which 7121 have public full texts. So, based on my quick investigation, it looks like that the part of Mendeley's social sharing site which deals with Computer Science (insert boilerplate text about disciplinary differences) seems to be functioning on a similar level to a dedicated departmental repository. They have more bibliographic records; we have more (just) records with public full texts.

The interesting contrast between the Mendeley approach and the repository approach is that the one starts with services on the researcher's desktop that are then used as the basis for offering open access, whereas the other starts with web-based open access that leads to desktop bibliographic tools. It might appear, if my partial and approximate study is anything to go by, that neither approach trumps the other in terms of open access outcomes.

The repository community should certainly embrace and work with services like Mendeley, but we should see them as complementary to our activities, not a replacement for them!