Elsevier svarer de europæiske biblioteksorganisationer om Text and Data mining

RE: OPEN LETTER IN RESPONSE TO THE REQUEST FOR ELSEVIER TO WITHDRAW ITS TEXT AND DATA MINING POLICY

Dear Colleagues,

I am writing in regard to your open letter asking Elsevier to withdraw its text and data mining
policy. We would like to reassure librarians that we haven’t introduced our TDM policy to undermine
your calls for copyright exceptions, but rather as the next natural step in the evolution of our
TDM services, which have been available since 2006 and have evolved continually over this time. We
understand that librarians will continue to lobby for exceptions, and while we disagree on whether
these are necessary, would call on all stakeholders to agree that it is important to provide
researchers with practical, workable TDM services now no matter the legal framework of the country
in which they are based.

We would also appreciate the opportunity to elaborate further here on why we think our policy is
fit for the purposes of the research community as well as Elsevier. I would like to first make
clear that our concern is – and has always been – to ensure that TDM works for researchers in
practice. We are not debating the merits of enabling TDM for scientists per se; that for us is a
given. The discussion is instead about how we ensure TDM works in practice and for the long term,
and we have tailored our services to support researchers globally regardless of the legal regime of
the country in which they work. We invite library organizations to actively support practical
initiatives such as ours to serve the text and data mining needs of researchers instead of focusing
on changes to copyright, which may provide the legal framework but not the necessary tools for TDM.

Elsevier’s text and data mining policy enables researchers to text and data mine content to which
their institution has legal access, whether via a license or an exception. Our policy applies
globally, even in the UK, where our policy and service provision align with the provisions in the
new copyright exception. Further details on all of the above can be found in a recent article we
published, which describes how Elsevier’s text and data mining policy works with the UK exception:
http://www.elsevier.com/connect/how-does-elseviers-text-mining-policy-work-with-new-uk-tdm-law

Elsevier has invested significantly to develop a specially-designed technological infrastructure to
facilitate mining and support researchers. For example, via an API, we provide full-text content in
XML and plain-text formats – the formats researchers have told us they prefer for mining. The
separate infrastructure we provide for miners also ensures we can continue to serve the wider
research community, as our reading platforms are not compromised by high volume downloading. Our
API is designed in accordance with web standards and protocols, and we do not require that any
specific tools be used to access it.

Your open letter raises a number of points that we would like to address in detail:

• Our Click-Through Agreement: For the avoidance of doubt, our click-through agreement with
researchers is not intended to place additional liability on the researcher, nor do we think it
does so. On the contrary, the click-through agreement serves merely to advise users of the terms
of use for our API. We take this step because we want to make sure that researchers conducting text
mining are aware of their rights and obligations under those terms. Similarly, we have no intention
to block or suspend the API or adjust the terms in ways that would negatively impact researchers
using it. We have included the provision which allows us to update the agreement from time to time
in order to ensure that we can adjust the terms and conditions of use of APIs in response to
community feedback without the undue administrative burden of requiring all researchers who have
previously accepted it to re-accept.

• The Role of the Institution as an Intermediary: Elsevier is in no way seeking to undermine the
role of the institution as intermediary or to restrict researchers’ privacy, contrary to the
concerns outlined in your open letter. Instead, our API registration process is there so that we
can support researchers; the issuing of a personalized API key allows us to provide efficient
customer support and to contact the right individual if an automated process using our APIs goes
wrong and impacts the service for other users. The issuing of an API key via our self- registration
portal is an entirely automatic process, and we in no way vet or review the information provided by
researchers prior to providing or activating the API key – the key may be used immediately. Also,
in response to researcher feedback, we no longer request a project description as part of the API
registration process.

• Mining Images: We have always sought – and will continue to do so – community feedback, and
make tweaks and changes to our policy where appropriate. As your letter notes, researchers often
want to mine images and other content, not just text. Although we have received very few requests
for the mining of images and have, since 2006, made this content available to researchers on
request, we now make images, graphs, tables and other non-textual content available for mining via
our API. It is not the case that we wish to restrict the content researchers can mine, but we,
too, are bound by legal obligations to respect third-party copyright and often do not hold reuse
rights to the images we license. Therefore, while TDM on images is allowed, the original image
cannot be substantially reproduced as part of the TDM output unless permission is obtained from the
image rightsholder. We will continue to monitor this area carefully.

• Negotiating Licenses with Individual Researchers: We agree with you that it would be
unfortunate to distract researchers from their research by requiring them to negotiate TDM licenses
with individual publishers. That is why we are participating in the CrossRef Text and Data Mining
service, which aims to allow researchers to mine content from a range of publishers through a
single common access point, and why we have aligned our click-through license terms to those in the
STM model license that was developed in consultation with librarians. Further information on this
service can be found here: http://tdmsupport.crossref.org/ and
http://www.stm-assoc.org/text-and-data-mining-stm-statement-sample-licence/

• Licenses on outputs: We support the right of researchers to share their research as widely as
possible, including the results of their text and data mining, and continue to make changes in
response to researcher feedback. Again, thanks to feedback from researchers, we no longer require a
specific license to be placed on outputs, but instead ask researchers to make clear by means of a
notice that the content they are displaying as an output may belong to others and

should only be re-used on a non-commercial basis provided the original author and source are
credited. As set out above, this does not currently apply to images, which cannot be displayed as
TDM output for the time being due to our existing legal obligations.

• Transparency and reproducibility of results: We absolutely support the principles of
transparency and reproducibility of results. To this end, we provide stable, well-maintained public
identifiers for each of our articles in the form of Digital Object Identifiers (DOIs). Researchers
may share and publish details of the dataset used for their research by referring to the DOIs
contained within the dataset, and other researchers may reproduce that research by accessing the
very same dataset via our APIs reusing those DOIs. We are also working in myriad other ways to make
data more transparent and accessible, for example through our endorsement of the Joint Declaration
of Data Citation Principles and through our linking agreements with a number of data repositories.
You can read about some of these initiatives here:
http://www.elsevier.com/connect/building-a-global-infrastructure-for-research-data

If you have specific concerns about details in our API agreement, we would be happy to hear from
you so that we can address them. In return, you may be interested in an upcoming workshop on text
and data mining, co-organised by Mendeley, CORE, the Open University, Athena Research and
Innovation Centre, and Europeana:
http://www.elsevier.com/connect/call-for-papers-international-workshop-on-
mining-scientific-publications.

Thank you again for the opportunity to explain our policy and why we think it is well suited to
support researchers. We recognize that you will continue to seek changes, and we look forward to
continued discussion on the subject.

Yours sincerely,

Gemma Hersh Policy Director [email protected]
@gemmahersh

10th July, 2014