Libraries at the centre of the debate on copyright and text and data mining: the LIBER experience by S K Reilly
1. Libraries at the centre of the debate on copyright and
text and data mining:
the LIBER experience
Susan Reilly
19th
August 2014
IFLA, Lyon
2. Text & Data Mining is the future
“Text and data mining (TDM) is the process of deriving
information from machine-read material. It works by
copying large quantities of material, extracting the data,
and recombining it to identify patterns.” JISC
3. Alternative to literature review
• Over 50 million articles online
• 1.5 million articles published annually
• Advanced discovery and visualisation
• “Undiscovered public knowledge” (Swanson)
4. Malhotra A, Younesi E, Gurulingappa H, Hofmann-Apitius M (2013) ‘HypothesisFinder:’ A Strategy for the
Detection of Speculative Statements in Scientific Text. PLoS Comput Biol 9(7): e1003117.
doi:10.1371/journal.pcbi.1003117
5. “TDM saves lives”
http://arxiv.org/abs/1407.7094
• Tools in the armoury of every biologist and
biotecnician
• Discover new treatments for diseases e.g. fish
oil for Raynaud’s Syndrome
• Controlling malaria outbreaks
• Links between gene mutation and cancers
7. Anyone can use TDM tools!
E.G. Sentiment analysis of IFLA open letter to European Union
8. Economics (Europe)
• TDM potentially worth 5.3 billion euro a year to European
research budget (2%)
• Knock-on effect would be a minimum of 32.5 billion euro
increase in GDP
• US responsible for over half
the articles and patents on TDM
- 1100 US patents compared to 39
EU by 2013
9. Copyright v TDM
• Because it involves the copying of content in
order to convert into machine readable format
TDM may infringe copyright
• European Database Directive
prohibits copying of substantial
parts of databases
• In US TDM is covered
by fair use, other parts of the
world have a specific exception
e.g. Japan, UK
https://www.flickr.com/photos/apelad/304195427/
10. The debate in Europe
• Licences for Europe, Feb 2013
– “The Commission's objective is to promote the efficient use of text and data
mining (TDM) for scientific research purposes. ……The Group should explore
solutions such as standard licensing models as well as technology platforms to
facilitate TDM access.”
• No discussion of copyright e.g. does TDM
infringe copyright law?
• Engaging the wrong stakeholders
• An attempt to systematise a problem/not a
solution
11. The problem with licences
• Permission culture: Why relicence? Can’t licence
everything!
• Not scalable or cost effective
• Will licence reflect how the researcher actually
performs TDM?
ME 442 Permission" by Nina Paley - http://mimiandeunice.com/2011/08/30/permission-2/. Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons -
http://commons.wikimedia.org/wiki/File:ME_442_Permission.png#mediaviewer/File:ME_442_Permission.png
12. So, we walked away…..
• We want to be free to mine content to which we
have legal access
• Copyright reform required
• Real stakeholder
engagement
13. The Perfect Swell: ideal conditions
for growth of TDM in Europe
• Stakeholder workshop (60 attendees)
• Views from industry, researchers, infrastructure, OA
publishers, legal experts
• Main findings:
– Licencing not scalable
– Need to address lack of legal clarity (does TDM
infringe copyright?)
– Need for harmonisation of copyright law
– Lack of awareness amongst researchers
– Publisher infrastructure not threatened by TDM
http://blogs.plos.org/opens/2014/03/09/best-practice-
enabling-content-mining/
14. So, what do we want?
• Legal clarity
– A specific exception in EU law to allow TDM
– A reinterpretation of EU law
• Legal interoperability
– A solution at WIPO
• Open licences
– CC-by and CC0
15. What do we not want?
• Licences for subscriptions which explicitly forbid
machine crawling
• A licence with every single publisher for every
single research project
• Publishers placing conditions on how TDM
results are disseminated
• Click-through licences
• “Open access” licences that are
NOT interoperable (STM model licences)
16. Elsevier TDM Policy
• Access through API only
• Text only- no images, tables
• Research must register details
• Click-through licence
• Terms can change any time
• Reproducibility of results
17. By MsSaraKelly (Back to the Future by Graffiti Life) [CC-BY-2.0
LIBER will:
•Advocate for copyright reform
in Europe
•Support international efforts
for harmonisation
•Continue to engage research
stakeholders
•Promote open access
Thank you.
As introduced, my name is Susan Reilly and I am Advocacy and Projects Manager for LIBER, the Association of European Research Libraries. LIBER represents over 400 research libraries (that is national, university, and other dedicated research libraries) in over 40 countries. Our mission is to create an information infrastructure to enable research in LIBER institutions to be world class.
In pursuit of this mission, since February 2013, LIBER has been advocating for copyright reform, both in Europe and internationally, in order to ensure legal clarity around the act of text and data mining and therefore increase the practice of it. We are advocating for reform because we believe that text and data mining will become integral to the research process, it will bring new efficiencies to research, increase analytic capability and provide new insights that would not be possible without machine technology. In short, we do not believe that, in today’s world of advanced analytics and increasing machine power in research, an information infrastructure can enable world class research outputs without a copyright framework that supports the way researchers work in the digital age.
I’m going to outline our trajectory in terms of how and why we started advocating for copyright reform, but first, in order to explain our position (and maybe bring a few more of you on-board), I’d like to take a look at the definition of text and data mining, what it involves, how this relates to copyright and what the current situation is in Europe.
Text and data mining (TDM) is the process of deriving information from machine-read material. It works by copying large quantities of material, extracting the data, and recombining it to identify patterns. TDM is essentially another method of reading, done by the computer rather than the human eye. It is a natural next step for the research process, as more and more content is electronic. For libraries what this means is that researchers are able to extract more value from our vast collections- born digital and digitised. I’d like to show you some examples of the added value of TDM.
TDM can act as an alternative to a literature review. In 2013 the figure for online articles was over 50 million and the rate of publication was projected to be 1.5 million a year. The large volume of research publications can make it impossible to conduct a comprehensive literature review but using TDM for advanced discovery can help narrow, or uncover patterns in the literature. In fact even library discovery services use TDM e.g. to visualise search results.
Don Swanson, an information scientist, first wrote about undiscovered public knowledge in 1986. He believe such knowledge was available by bringing diverse literature together- in this way he made a connection between dietary fish oil and Raynaud’s disease, a circulatory disorder.
A nice example of how TDM can replace or supplement a literature review is this hypothesis finder. An hypothesis is a supposition or proposed explanation made on the basis of limited evidence as a starting point for further investigation.
It’s now widely agreed that text and data mining tools should be in common use by biologists and biotechnician. It’s been proven that the mining of sceintfic literature can provide new insight into the treatment for diseases, it can identify where medicines that are already on the market can be used to treat different diseases- dramatically reducing trial times. Data can also be mined to identify the spread of outbreaks and predict and hopefully prevent violent events.
TDM can provide us with new cultural insights. Take this recent visualisation of the geographic dispersal of Spanish dialects which was created by mining geo located tweets. The research has uncovered super dialects and where they are being used, whether the areas are rural or urban. Although the research itself was carried out by European researchers, the datasets were analysed by a company in the US, probably because of the lack of clarity around performing the same activity in Europe.
TDM is not just for developers. We can all use TDM tools and they will eventually lead to more government transparency, for example. Just to prove that TDM is for everyone, here’s an example. This is a tool from the National Centre of text and data mining in the UK. It’s in beta, but is freely available and can be used to analyse the sentiment of a text. In this example I have taken a letter that the IFLA CLM Committee recently sent to the European Commission. The letter was sent on foot of a setback at WIPO in relation to the work on copyright exceptions for libraries. This setback was caused by the EU. This letter demonstrates what pros the people at IFLA are when it comes to letter writing. Note, the only reason I could perform this analysis and not break European copyright law is because the content on the IFLA Website is available under a cc-by licence.
BL estimates 16 months to negotiate a new licence
Publisher expectation that each researcher will provide details of research project unrealistic
Additional licence for TDM unecessary
Storing the data
Sharing the data