PreprintPDF Available

Limitations to Text and Data Mining and Consumer Empowerment Making the Case for a Right to "Machine Legibility"

Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

The paper focuses on the current legal barriers to text and data mining (TDM) in the context of smart disclosure systems (SDSs) whose aim is to provide consumers with improved access to the data needed to make informed decisions. The use of intellectual property rights and contracts, combined with technological protection measures, can hinder TDM and the deployment of SDSs. Further, those legal constraints can negatively impact artificial intelligence innovation that requires improved access to data. There are thus various arguments for enhanced “machine legibility”. However, the TDM exception included in the draft Copyright in the DSM Directive and the various amendments proposed by the European Parliament or the Council do not appear to clear the way for enhanced “machine legibility”. In relation to SDSs, we also argue that the principle of transparency, embedded in consumer and data protection laws, can serve as a last line of defence against prohibition of TDM.
Content may be subject to copyright.
CRIDES Working Paper Series
31 October 2018
by Rossana Ducato and Alain Strowel
Cite as: R. Ducato, A. Strowel, Limitations to Text and Data Mining and
Consumer Empowerment: Making the Case for a Right to “Machine
Legibility”, CRIDES Working Paper Series, 31 October 2018
Limitations to Text and Data Mining and
Consumer Empowerment
Making the Case for a Right to “Machine
The CRIDES, Centre de recherche interdisciplinaire Droit Entreprise et Socié, aims at investigating,
on the one hand, the role of law in the enterprise and, on the other hand, the function of the
enterprise within society. The centre, based at the Faculty of Law UCLouvain, is formed by
four research groups: the research group in economic law, the research group in intellectual
property law, the research group in social law (Atelier SociAL), and the research group in tax
This paper © Rossana Ducato and Alain Strowel 2018 is licensed under a
Creative Commons Attribution-ShareAlike 4.0 International License
Rossana Ducato* and Alain Strowel
The paper focuses on the current legal barriers to text and data mining (TDM) in the context of
smart disclosure systems (SDSs) whose aim is to provide consumers with improved access to the
data needed to make informed decisions. The use of intellectual property rights and contracts,
combined with technological protection measures, can hinder TDM and the deployment of SDSs.
Further, those legal constraints can negatively impact artificial intelligence innovation that requires
improved access to data. There are thus various arguments for enhanced “machine legibility”.
However, the TDM exception included in the draft Copyright in the DSM Directive and the various
amendments proposed by the European Parliament or the Council do not appear to clear the way
for enhanced “machine legibility”. In relation to SDSs, we also argue that the principle of
transparency, embedded in consumer and data protection laws, can serve as a last line of defence
against prohibition of TDM.
Text and data mining Copyright Sui generis right Exceptions - Consumer protection Data
protection Smart disclosure systems
* Rossana Ducato ( is a postdoctoral researcher in Law at UCLouvain and
Université Saint- Louis Bruxelles.
The research is supported by the Innoviris grant 2016-BB2B-9. A sincere thanks to dr. Guido Noto La
Diega for our constructive discussion on an early draft of this paper.
Alain Strowel ( is professor at UCLouvain, Université Saint-Louis
Bruxelles, KULeuven, Munich IP Law Centre.
CRIDES Working Paper Series 2018
1. Introduction
Smart disclosure refers “to the timely release of complex information and data in
standardized, machine readable formats in ways that enable consumers to make informed
decisions”.1 Smart disclosure systems (SDSs) allow users to get easy and timely access to the
relevant pre-contractual information or even receive personalised advices based on their
preferences.2 Over the last two years, several initiatives have emerged worldwide offering third
party services to automatically analyse websites’ contractual documents and to check their
compliance with applicable consumer and data protection laws.3 One of the main goals of these
projects4 is to increase the awareness of users towards the rights, obligations and possible risks in
their online transactions, trying to reduce or overcome the well-known signing-without-reading
problem.5 Such tools are primarily directed to consumers, as end-users of the service; however,
other possible users are consumer associations or regulatory authorities, which can use them to
perform periodical assessments, start investigations or verify complaints more quickly.
The functioning of SDSs is based on text and data mining (TDM). TDM uses techniques
from natural language processing, machine learning, information retrieval, and knowledge
1 Sunstein 2012.
2 On the advantages of smart disclosures and targeted information, see Ben-Shahar 2009; Helberger 2013;
Porat and Strahilevitz 2013; Bar-Gill 2015; Busch 2016; Helleringer and Sibony 2017; Busch forthcoming.
3 One of the first projects in the area of automated analysis of legal documents is “Usable Privacy Policy”
(, a consortium led by Carnegie Mellon University. Their tool aims to help users to
navigate through the text of privacy policy and identify the privacy options and choices available (Sadeh et al.
2013). More recently, an international team formed by researchers from the Switzerland's Federal Institute of
Technology, the University of Wisconsin and the University of Michigan, has launched two tools: Polisis
(, a tool to visualise in a very effective way the content of a privacy policy, and Pribot
(, a chatbot available to answer questions about a specific privacy policy (Harkous et al.
2018). In the field of the automated analysis of T&C, we must mention CLAUDETTE, a research project
carried out by an interdisciplinary team at the European University Institute ( The
tool, based on machine learning techniques, assesses the fairness of consumer standard terms
( This functionality will be extended to the analysis of privacy policy.
For more information, see Contissa et al. 2018; Lippi et al. 2018. Another interdisciplinary project, SaToS
(Software Aided Analysis of ToS), is conducted by the chair of Software Engineering for Business Information
Systems (Sebis) at TU Munich. The German research group is developing a solution to automatically identify
Terms of Services of e-commerce websites and summarise the key points of the contract in a simplified language
(Braun et al. 2018).
4 This is precisely one of the objects of “The Internet of Platforms: an empirical research on private ordering
and consumer protection in the sharing economy”, carried out at UCLouvain. The project aims to address the
issue of the lack of transparency in sharing economy transactions and improve the information users receive
from and about the platform ( This paper presents some
of the preliminary results of this project.
5 On the users’ tendency not to read contracts, the literature is extensive. Ex multis, Vila et al. 2003;
Wilhelmsson 2004; Hillman 2006; Ben-Shahar 2013; Radin 2013; Ayres and Schwartz 2014; Bakos et al. 2014.
Limitations to text and data mining and consumer empowerment
management for the automated analysis of digital content (structured and unstructured data), in
order to extract information, identify patterns, discover new trends, insights or correlations.6
Despite the development of such promising tools for enhancing the readability and
understandability of the conundrum of terms, the current EU legal framework is not particularly
supportive of TDM. Many rigorous studies have already analysed the barriers to TDM, current
tensions and negative externalities in the context of research, contributing to the debate on the
TDM exception in the ongoing copyright reform. 7 Several of these studies highlight that the
beneficial uses of TDM are not limited to scientific research, but take place in other contexts,
including consumer information and protection.
In the light of the artificial intelligence (AI) applications that are emerging, it is easy to
predict that TDM will become a central technique enabling new kinds of information-based
services and applications (e.g., for care and medical purposes, fact checking, disaster prevention,
elaboration of sustainability policies and politics, etc.). Many studies have already highlighted the
need for a broad access to datasets so as to train algorithms and improve AI applications.8
The article aims to contribute to the discussion on TDM by adding a further perspective.
We take into consideration the application of TDM in SDSs, a sector that is growing in
importance. The automated analysis of contracts and privacy policy for enhancing the awareness
of consumers and, ultimately, ensuring consumer empowerment, is a perfect lab to test TDM’s
pierres d'achoppement.
The article is structured as follows: following the present Introduction, we outline in
Section 2 a taxonomy of the possible obstacles for TDM and SDSs, identifying intellectual
property rights (IPRs), contracts and technological protection measures (TPMs) as the principal
In Section 3, we analyse the interplay between copyright, the database sui generis right
and what is called private ordering9, noting that the existing copyright limitations may offer little
help to counterbalance the power of contracts.
6 In the proposal for a Directive on copyright in the Digital Single Market (hereafter “Copyright in the DSM
Directive”), TDM is defined as “any automated analytical technique aiming to analyse text and data in digital
form in order to generate information such as patterns, trends and correlations” (Art. 2.2, Proposal for a
Directive of the European Parliament and of the Council on copyright in the Digital Single Market,
COM/2016/0593 final - 2016/0280 (COD)). The definition is sufficiently broad to embrace the current TDM
application panorama. For a technical definition of text and data mining, see Hearst 2003. Specifically on text
mining, Feldman and Sanger 2007. For an extensive analysis of the definition of TDM, see Triaille et al. 2014.
7 Ibid.; Triaille et al. 2014; Bernhardt et al. 2015; Caspers and Guibault 2016a; Margoni and Dore 2016;
Stamatoudi 2016; Hilty and Richter 2017; Geiger et al. 2018; Margoni and Kretschmer 2018; Rosati 2018.
8 On the crucial need to train algorithms on different datasets, see Hall and Pesenti 2017.
9 By “private ordering”, we refer to both contractual, technological and informal measures as tools to enforce
platforms’ rights and interests towards their users. In the absence of a clear legislative framework or effective
(and efficient) remedies, contracts and technology can be used to expand the prerogatives and powers of
CRIDES Working Paper Series 2018
Therefore, in Section 4 we explore whether the current proposal for a specific TDM
exception in the Copyright in the DSM Directive10 and the ongoing copyright reform in the EU
will fill some of these gaps. Despite some merit of the text proposed by the European
Commission (and of the amendments of the European Parliament), the exception as drafted will
not permit to fully embrace the potentialities of Big Data analytics and AI, nor will it dispel legal
uncertainty, as contemplated by the European Commission.
In Section 5 we go further and argue that the TDM and related data access issue should
be framed in a perspective that takes copyright’s rationale seriously. This leads to the argument
that the reproduction right should in the first place not cover TDM processes.
The limitations to TDM coming from private ordering will be examined in Section 6,
where we present the results of an empirical analysis conducted on a representative set of online
platforms operating in the sharing economy. This analysis shows that there is a widespread trend
to preclude TDM via contracts (the online Terms & Conditions) and through the embedded
code. A prohibition of TDM making it impossible to run tools for automated contractual analysis
would prevent consumers to access the relevant information to make informed choices. This
would not be in line with the principle of transparency enshrined in both consumer and data
In Section 7 we conclude by presenting the principle of “machine legibility” which
transposes the transparency principle in the technological context.
2. Obstacles to TDM in the current European framework: setting the scenario for smart
disclosure systems
There are at least three legal protections and tools that could limit the practice of TDM:
1) intellectual property rights (IPRs); 2) contracts; 3) data protection rules. First, at the light of
the “black letter” of copyright law and its current interpretation by the Court of Justice of the
EU (ECJ), many TDM activities could be considered as copyright infringements or violations of
the database sui generis right. 11 Second (and this is an equally worrisome signal in the never-
ending battle for the control of information), contracts may expressly prohibit or limit TDM. In
addition, TPMs can enforce (and reinforce) IPRs and contractual provisions, impeding TDM in
practice. Finally, if the object of the mining consists of personal data, i.e. “any information
platforms, restricting the legitimate uses and faculties of the weaker party. As noted with reference to the
intellectual property domain by Dussolier 2007, p. 1393-1394.
10 Art. 3, Proposal for Copyright in the DSM Directive and the amendments presently discussed in the
European Parliament.
11Triaille et al. 2014; Margoni and Kretschmer 2018; Stamatoudi 2016; Montagnani and Aime 2017; Strowel
Limitations to text and data mining and consumer empowerment
relating to an identified or identifiable natural person”12, the processing has to be compliant with
data protection law.
These three legal instruments have different scopes and objectives but, in some cases,
they produce the same consequences for TDM. Imagine a mining activity carried out by an
insurance company on a national electronic health record system: TDM would be forbidden and
several legal instruments could be invoked to justify it, in particular the protection of personal
data. In other cases, data protection and copyright could admit the mining for research purposes
(as both protections are subject to limitations for research), but TPMs, impeding the bulk
download of the content, could affect the conduct of research in practice. Meanwhile, some
mandatory copyright exceptions and lawful uses cannot be overridden by contractual provisions.
It is therefore a complex and dynamic scenario that needs to be further explored if we want to
unleash the potential of SDSs and, beyond, AI applications.
In the case of SDSs, the mining covers legal documents, such as Terms & Conditions
(T&C) and privacy policy. Therefore, the issue of data protection will be left out from the
analysis. Furthermore, for the purpose of this paper, the focus will be on IP and contractual
Fig. 1. Representation of limits to TDM in the context of smart disclosure systems.
12 Art. 4(1), General Data Protection Regulation.
CRIDES Working Paper Series 2018
3. Before the (IP) Law: limits and counter-limits to TDM
IP is one of the first barriers to TDM. The latter may, in principle, clashes with a bundle
of exclusive rights if the work or the database qualify for protection under Directive 2001/29/EC
(“InfoSoc Directive”) or Directive 96/9/EC (“Database Directive”).
As known, copyright protects “databases which, by reason of the selection or arrangement
of their contents, constitute the author’s own intellectual creation”.13 A database is defined as a
collection of independent works, data or other materials arranged in a systematic or methodical
way and individually accessible by electronic or other means”.14 A website, like that through
which online platforms offer their service (for example, the AirBnB website), can certainly fall
into this notion.15 Copyright covers the database’s peculiar “expression”, i.e. the originality of its
systematic organization, which is exteriorised through the “free and creative choices”16 that show
the “personal touch”17 of the author. A contrario, the requirement of originality is not satisfied
when the setting up of the database is dictated by technical considerations, rules or constraints
which leave no room for creative freedom”.18 Therefore, copyright does not apply when the
selection and arrangement follow a chronological or alphabetical ordering.19
Most websites of the sharing economy platforms that can be considered as databases will
probably not reach the threshold of originality required for copyright to protect the database as
such (to be distinguished from the content of its pages). First of all, because the selection and
arrangement of the data is essentially shaped by the kind of service that the platform provides. A
carpooling platform will allow users to search for the closest car available, displaying, for example,
the distance, the location of the car, the level of fuel, the license plate. In addition, to be user-
friendly and easy to retrieve, some information has to be organised according to “trivial” criteria:
e.g. T&C, privacy policy, copyright notices, FAQs, norms of the community are generally
included in the “Legal conditions” section. Therefore, in the given scenario, copyright is unlikely
to apply to the platform’s database.
However, the database can be protected under the sui generis right regime. The latter is
an exclusive protection granted to the maker of the database which shows that there has been
qualitatively and/or quantitatively a substantial investment in either the obtaining, verification
13 Art. 3, Database Directive. For a comprehensive overview, see, Beunen 2007; Derclaye 2008, 2014.
14 Art. 1(2), Database Directive.
15 For the classification of a website as a database, see Strowel and Derclaye 2001, p. 311-312.
16 ECJ, Case C-604/10, Football Dataco Ltd and Others v Yahoo! UK Ltd and Others [2012],
ECLI:EU:C:2012:115, para. 38. Derclaye 2012; Rosati 2013b.
17 ECJ, Case C-604/10, Football Dataco Ltd, para. 38.
18 Ibid., para. 39.
19 In line with the US leading case Feist Publications Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991). See,
Waelde et al. 2013, p. 65.
Limitations to text and data mining and consumer empowerment
or presentation of the contents”.20 We could debate whether there is substantial investment in
obtaining data, if the latter are constantly produced by the web-users (e.g., personal information,
localisation data, pictures, preferences). Equally, one could wonder whether there is substantial
investment in the verification of the content if the platforms or sharing economy websites
explicitly state and follow the policy that they do not check the accuracy of the data provided by
users. At the same time, there are rooms to allege that a substantial investment could lie in the
presentation of the content: after all, the platform arranges the website to facilitate the sharing
of information between the users, adjusts the filters and search tools to customise and make
appealing the experience for the end-user.
If we assume for the sake of argument that the threshold of substantial investment is met,
the way the sui generis right is conceived and designed could affect the behaviour of the
legitimate user. Like copyright, the sui generis right does not need any formality to exist.21 At the
same time, the substantial investment in either one of the three activities listed in Art. 7
(Directive 96/9/EC), as interpreted by the ECJ22, is a determination that is likely to be verified
in court only and, considering the requested proof, the evidence is in the hands of the maker of
the database. This means that, unlike copyright, it will be difficult for a user to know, by simply
consulting the database, whether the latter is protected by the sui generis right.23 Hence, the sui
generis right framework can contribute to legal uncertainty, reducing in practice users’ faculties
that would be totally legitimate: a user who wants to avoid the risk of private sanctions (the
suspension of the account or the ban from the platform) or legal actions, will adopt a
precautionary approach. She will act as if the sui generis right protects the database (by the way,
the standard T&C will confirm that the provider of the online platform intends to protect the
website by IPRs).
Another issue is whether copyright protection applies to the T&C and/or the privacy
policy, which will be primarily investigated in the case of automated analysis of contractual
documents. There is some discussion as to the copyrightability of such documents. In a way, they
are “standard” by definition and their formulas are usually based on templates. Furthermore, in
the case of privacy policy, its structure is mainly determined by law (see, for instance the list of
mandated disclosures in Arts. 13 and 14 of the General Data Protection Regulation, “GDPR”).
However, the EU criterion of originality does not require novelty or a high level of creativity. It
20 Art. 7, Database Directive.
21 Derclaye 2008, p. 107.
22 ECJ, C-46/02, Fixtures Marketing Ltd v. Oy Veikkaus Ab [2004] ECLI: ECLI:EU:C:2004:694; C-338/02,
Fixtures Marketing Ltd v. Svenska Spel AB [2004] ECLI:EU:C:2004:696; C-444/02, Fixtures Marketing Ltd v.
Organismoa Prognostikon Agnon Podosfairou AE (OPAP) [2004] ECLI:EU:C:2004:697; C-203/02, The British
Horseracing Board Ltd and Others/William Hill Organization Ltd [2004] ECLI:EU:C:2004:695.
23 The maker of the database could state the existence of the sui generis right in the T&C, but such a
circumstance is quite rare. Usually, T&C contain general formula such as “All right reserved”, leaving ample
room for interpretation to the end-user, which in the majority of the cases is nor a lawyer, much less an IP
CRIDES Working Paper Series 2018
suffices to demonstrate the “author’s own intellectual creation”24, which can be rather modest.25
Often there is still room for choices and adjustments in the presentation and drafting of those
documents, which means they could be protected by copyright.
There is no case law concerning the standard of originality applied to contractual texts.
Therefore, it is necessary to refer to the general framework and to reason in consimili casu. For
instance, in Infopaq I, the Court of Justice stated that words, as such, are not protectable, but
through the choice, sequence and combination of those words […] the author may express his
creativity in an original manner and achieve a result which is an intellectual creation”26. Thus,
the ECJ concluded that even eleven consecutive words can potentially “express the author’s own
intellectual creation27, leaving such a determination to national courts. The latter have ruled,
episodically, over similar issues, i.e. the creativity of legal or technical works. In the Italian case
law, for example, a technical article (a sort of machine’s instruction manual) describing the
functions of a monoscope has been considered original.28 In another case dealing with legal
guidelines, the District Court of Venice found that “a regulation against counterfeiting” written
by a lawyer presented the degree of originality required by copyright.29 In Spain, the Madrid
Provincial Court concluded the same in relation to an exercise book of mathematical problems
in statistics.30 Under Belgian and French copyright laws, instruction manuals and other
informational documents have been recognized as protected by copyright.31 For instance, the text
of a patent before its official publication32 as well as the wording of various contracts33 have been
24 As known, the originality requirement is not harmonised in the InfoSoc Directive (it is mentioned in the
Software, Term and Database directives). However, the concept has been interpreted as “the author’s own
intellectual creation” and applied to works in several decisions of the European Court of Justice. See, ECJ, C-
5/08, Infopaq International [2009] ECLI:EU:C:2009:465, para. 37; C-403/08, Football Association Premier League
and Others [2011] ECLI:EU:C:2011:631, para. 97; C-393/09, BSA [2010], ECLI:EU:C:2010:816, paras. 44-45;
C-355/12, Nintendo and Others [2014], ECLI:EU:C:2014:25, paras. 21-22. For an overview of the
“Europeanization” of originality, Strowel 2012.
25 Rosati 2013a. See also ECJ, Case C-161/17, Land Nordrhein-Westfalen v Dirk Renckhoff [2017]
ECLI:EU:C:2018:634 (on a photograph of the city of Cordoba).
26 ECJ, Case C-5/08 Infopaq International, para. 45.
27 Ibid., para. 48.
28 District Court of Milan, specialised section, decision no. 6057/2014, available online here:
29 District Court of Venice, specialised section, decision of 17 December 2014, RG 1522/2011 (not
30 Madrid Provincial Court (s. 12) of 3 march 2004, cited in Vallés 2009, pp. 114-115.
31 See Bernault et al. 2017, p. 116, footnote 94.
32 Paris Criminal Court, 17 January 1968, Gaz. Pal. 1968, I, p. 197.
33 Paris Commercial Court, 67ème ch., 4 Sept. 1989, Expertises, 1991, p. 273, obs. Gross (contract proposed
by a credit provider to traders); Paris Court of Appeal, 4th ch., 27 Nov. 2002, Expertises, 2003, p. 190 (software
licence); Douai Court of Appeal, 27 March 2013, Propr. Intell., 2013, p. 285, obs. Brugière (terms of reference
for public procurement).
Limitations to text and data mining and consumer empowerment
protected in France. In Belgium, the instructions for using IT equipment34 have been granted
copyright protection. In the UK, a wide range of subject matter has been protected as
compilations in the past, including a leaflet conferring information about herbicides.35 Applying
a low threshold of originality, courts have accepted as original railway tables and exam papers.36
In Germany, various decisions of the Federal Supreme Court (BGH) have considered as
protected by copyright the following technical documents: user guidelines for the usage of a
technical apparatus37, technical rules to be applied in the construction of roads38, technical
Therefore, the outcome of the intellectual effort of the author in drafting the various
clauses of some online T&C, the choice of words or the structuring of the document, could
qualify as protected work. Who has ever written a complex contract knows that it can be a
creative task, and that many free choices have to be made for the organisation of the clauses and
the drafting of each. In addition, thanks to some legislative interventions and a few recent
scandals in the field of consumer and data protection40, there is a growing trend towards the
encouragement of user-friendliness in legal documents: e-commerce sites, social networks and
online platforms in general are starting to provide T&C and privacy policy in novel formats,
using different fonts, layout and icons to increase the understandability, transforming the
“legalese” in a plainer language, etc. The drafters of the text of the creative commons licenses claim
they are protected by copyright, and made them available under the CC0 Public Domain
Dedication.41 In exceptional cases, designers are even involved in the drafting, this circumstance
making it hard to contest the originality of the legal documents.42
To sum up, to retrieve the relevant information, a SDS can automatically analyse
contracts and privacy policies, considered as original works, and the various legal sections of a
website could be considered as part of a database, potentially protected by the sui generis right.
34 Brussels Court of Appeal, 28 January 1997, cited in de Visscher and Michaux 2000, p. 31, footnote 125.
35 Supreme Court of Judicator Court of Appeal, Elanco v. Mandops [1980] RPC 213.
36 Bently and Sherman 2009, p. 64.
37 Federal Supreme Court, 10 October 1991 - I ZR 147/89 (“Bedienungsanweisung”).
38 Federal Supreme Court, 11 April 2002 - I ZR 231/99 (“Technische Lieferbedingungen”).
39 Federal Supreme Court, 22 September 1999 - I ZR 48/97 (“Planungsmappe”).
40 The speech of Senator Kennedy at the Congress hearing of Mark Zuckerberg for the Cambridge Analytica
affair has become viral: “Here's what everybody's been trying to tell you today, and and I say this gently. Your
user agreement sucks […] The purpose of that user agreement is to cover Facebook's rear end. It's not to inform
your users about their rights. Now, you know that and I know that. I'm going to suggest to you that you go
back home and rewrite it. And tell your $1,200 an hour lawyers, no disrespect. They're good. But but tell
them you want it written in English and non-Swahili, so the average American can understand it. That would
be a start”.
41 See, point 5 of the Creative Commons Terms, .
42 See, for instance, the privacy policy elaborated by the designer Stefania Passera:
CRIDES Working Paper Series 2018
In case of IP protection, the TDM activity might come into conflict with the various
exclusive rights included in each IP bundle. Depending on the technique used, TDM may
- the reproduction and/or communication to the public of a content (the text of
T&C/privacy policy);43
- the extraction and/or reuse of a substantial part of the database.44
A communication to the public or a reuse do not always occur in the case of TDM: the
latter usually elaborates the information and publishes the results of the analysis in the form of
aggregate data, statistics, reports, etc. Therefore, unless the output of the TDM shows the whole
or the excerpts of the protected work or the database, there will be no communication to the
public or reuse.45
The most problematic issue with TDM is the broadness of the notions of reproduction
(for copyright) and extraction (for the database right): when the tool has to run the analysis, it
copies all or part of the work, it transfers all or a substantial part of the contents of a database to
another medium or it technically adapts or translates the content (e.g. conversion from a PDF
to another format). 46 So, these operations that are necessary steps in the TDM process in
principle fall under copyright or under the database right.
The right of reproduction belongs to the copyright owner, while the right of extraction
(even a temporary one, like the visualisation on a computer’s screen) of the whole or substantial
part of a database is granted to its maker. This means that the user cannot perform TDM without
the authorisation of the right holder or in the absence of a copyright exception.
The interplay between IPRs, contracts and TPMs is visible in the field of copyright
exceptions and limitations. Under the current EU framework (mainly defined by the InfoSoc
Directive), some exceptions could apply to TDM: for works protected under copyright, for
example, Art. 5(1) of the InfoSoc Directive allows for the temporary reproduction of works,
which are transient or incidental [and] an integral and essential part of a technological process
and whose sole purpose is to enable: (a) a transmission in a network between third parties by an
intermediary, or (b) a lawful use of a work or other subject-matter to be made, and which have
no independent economic significance”; Art. 5(2)(b) admits the reproduction for personal use;
43 We do not consider the distribution right here, because it applies when the work is incorporated in a tangible
article only (see, recital 28, Directive 2001/29/EC), which is not our case.
44 TDM may potentially perform also the reproduction and/or adaptation and/or communication and/or
distribution of the database itself. However, we do not analyse this case here.
45 In the same sense, Caspers and Guibault 2016a; Triaille et al. 2014.
46 Triaille et al. 2014, p. 32. According to the authors, this is not the case if the tool simply spots one or two
words through the text without making a copy of the work (e.g., spotting and counting the occurrences of the
word “malaria”). In particular, p. 31. Similarly, Stamatoudi 2016, p. 1261; Montagnani and Aime 2017, pp.
379 ff.
Limitations to text and data mining and consumer empowerment
and Art. 5(3)(a) includes a specific exception for teaching and scientific research. A similar
exception for research is included in the Database Directive.47 The maker of the database cannot
prevent the legitimate user from extracting insubstantial parts (qualitatively or quantitatively
considered), as long as she does not perform acts which conflict with normal exploitation of the
database or unreasonably prejudice the legitimate interests of the maker of the database”48 and
“cause prejudice to the holder of a copyright or related right in respect of the works or subject
matter contained in the database”.49
However, such exceptions may not confer effective rights to the consumers: they are
narrow, do not cover all the spectrum of TDM technologies, and are differently implemented in
Member States. Besides, the framework of rights and exceptions under the InfoSoc Directive and
the Database Directive is not completely homogeneous: what would be allowed under copyright
is not necessarily allowed under the sui generis right (and vice versa).
The research exceptions, for example, are limited for the (sole)50 purpose of illustration
for teaching or scientific research. 51 TDM would be permitted under these exceptions in very
few cases, e.g. in the context of non-profit scientific projects dealing with SDSs or for educational
purposes in order to demonstrate the functioning of the tool. Such exception arguably could
also cover the “training” of machine learning systems, on the model of human teaching, but such
use would nevertheless require to be justified by the non-commercial purpose to be achieved
through TDM and the indication of the source.52
The exception for private copying will allow even more limited uses: as reported in
Caspers and Guibault (2016), national laws have been quite restrictive in the implementation of
47 We do not take into consideration Art. 9 (a) Database Directive, since the exception for personal use applies
only to non-electronic database, which do not permit TDM in any case.
48 Art. 8(2), Database Directive.
49 Art. 8(3), Database Directive.
50 As required by Art. 5(3)(a) InfoSoc Directive and Art. 6(1)(a) Database Directive, but not by Art. 9 (b)
Database Directive. See, Derclaye 2008, p. 131.
51 For a complete analysis of the activities that could fall within the exception, Triaille et al. 2013, p. 359 ff.;
Guibault et al. 2012, p. 49 ff.; Montagnani and Aime 2017, pp. 385 ff.
52 See Arts. 5(3)(a) InfoSoc Directive and Arts. 6(2)(b) and 9(1)(b) Database Directive. Another inconsistency
between the two Directives must be noted with regard to the research exception. Under the Database Directive,
the reference to the source appears to be a mandatory requirement, while the InfoSoc admits the possibility
of not indicating the author’s names if “this turns out to be impossible” (Art. 5.3.a. InfoSoc Directive).
According to some authors, this difference is more a declamation than a substantial matter, considering the
general principle of “ad impossibilia nemo tenetur”, which will be applicable to the exception for database in
any case (see Montagnani and Aime 2017, p. 387, citing Walter and Von Lewinski 2010). Other authors
interpret literally the provisions (cf. Triaille et al. 2014, p. 70, as reported in Montagnani and Aime 2017, p.
CRIDES Working Paper Series 2018
such a copyright limitation, focussing on personal use, study or (small scale) research” 53, and
“sometimes the scope is limited to a few copies”54. Considering for example Italian law, the
reproduction for private use is legitimate with reference to printed works and as long as made
manually or with means that do not allow the distribution of the work (Art. 68, Law 633/1941).55
According to the majority opinion in the literature, the exception refers to the reproduction used
in the family circle.56 The exception could be extended to online works if the three steps test is
respected (Art. 71-nonies, Law 633/1941), but it has received a narrow interpretation by national
courts so far.57 Therefore, the Italian private copying exception is rather limited in the digital
environment and may not likely permit TDM performed by an individual or, a fortiori, by a
consumer association.
Furthermore, to rely on those exceptions of the InfoSoc Directive is no guarantee as both
the research exception and the private copying exception are optional and not imperative:
Member States are free to adopt them and, where they exist, the exceptions can be overridden
by contracts or TPMs.58
A stronger defence for TDM could, in principle, come from Art. 5(1) of the InfoSoc
Directive. In fact, the temporary reproduction is a mandatory exception that must be
implemented by all Member States. However, there are some drawbacks. First, mandatory does
not necessarily mean that the exception is imperative (non-overridable by contracts). When the
European Legislator wanted to limit the freedom of contracts, it expressly did so (see, for
instance, Art. 15, Database Directive, and Art. 8, Directive 2009/24/EEC, “Software
Directive”).59 There is nothing similar in the InfoSoc Directive that would prevent the
53 Caspers and Guibault 2016a, p. 34. See, also Helberger and Hugenholtz 2007. The private copying exception
has traditionally received little attention in the literature and the case law, apart from the issues related to the
copyright levies and fair compensation. See, Strowel 2015.
54 Caspers and Guibault 2016a, p. 34.
55 Valenti 2007b, p. 195.
56 Ibid. p. 202, para. III and the bibliography thereby cited.
57 Despite the potential “open-ended” nature of the three steps test. As reported by Margoni 2012. See also
Hilty et al. 2008. While, for phonograms and videograms, there is a specific provision: the reproduction is
permitted if done by a physical person solely for personal use and for non-commercial purposes, in compliance
with the applicable TPMs (Art. 71-sexies (1), Law 633/1941). The exception will not apply if the reproduction
is done by a third party (Art. 71-sexies (2), Law 633/1941) and if the works are available on-demand and
protected by TPMs or contracts (Art. 71-sexies (3), Law 633/1941). On the limits of the private copying
exception for the digital context and the interplay with TPMs, cf. Caso 2004; Montagnani 2007; Mazziotti
58 See Derclaye and Favale 2010; Helberger et al. 2013; Triaille et al. 2013; Caspers and Guibault 2016a.
Regarding TPMs, Art. 6(4) does not require that the Member States take appropriate measures to ensure that
the beneficiaries of the private copying exception do in practice benefit of the exception.
59 Derclaye and Favale 2010, p. 90.
Limitations to text and data mining and consumer empowerment
prohibition of such use by contract.60 Nevertheless, some countries, like Belgium, Ireland and
Portugal, have expressly excluded the possibility to overcome the temporary reproduction
exception via contractual means.61 The situation is not crystal clear in other Member States.62
A second (and more preclusive) problem concerns the content of the exception: Art. 5(1)
is crafted for caching and browsing activity63, but the cumulative conditions set out in that
provision, as restrictively interpreted by the ECJ64, hardly apply in TDM activities. In particular,
the copies made through TDM are not necessarily temporary, transient or accessory.65 It is also
far from clear that the “sole purpose” of this reproduction is “to enable a transmission in a
network between third parties”. Furthermore, a smart disclosure system does not necessarily meet
the independent economic significance element. The latter not only requires that the temporary
reproduction does not have to generate “an additional profit, going beyond that derived from
lawful use of the protected” 66 item, but the reproduction also does not have to “lead to a
modification of that work”67. Literally interpreted, this last requirement will exclude most of
TDM activities, since the data analysis process usually implies a transformation of the original
work for making it processable by the machine (e.g., the conversion from a format to another
one).68 Therefore, there are several limitations of the temporary copy exception that make it
irrelevant for exempting TDM.
Lastly, one has to consider whether TDM could fall in the lawful uses recognised by the
Database Directive at Art. 6(1) and 8. These rights are expressly protected against conflicting
contractual provisions (Art. 15, Database Directive). Art. 6(1) allows the lawful user to make a
copy of the database in order to access the contents or to allow a “normal use” of the same. It
would be coherent with the rationale of the Directive to consider a SDS aiming to analyse the
terms or the privacy policy as performing a normal use of the database.69 Meanwhile, with
60 The conflict between freedom of contracts and copyright limitations is extensively investigated in Guibault
2002. Even if with reference to the proposal for the InfoSoc Directive, the issue was already pointed out by
Hugenholtz 2000.
61 Kretschmer et al. 2010, p. 13.
62 In Italy, see Art. 68-bis, Law 633/41. Cf. Valenti 2007a.
63 Recital 33, InfoSoc Directive.
64 ECJ, C-5/08, Infopaq International A/S v Danske Dagblades Forening [2009], ECLI:EU:C:2009:465; C-403/08,
Football Association Premier League Ltd and others v QC Leisure [2011], ECLI:EU:C:2011:631; case C-302/10,
Infopaq International A/S v Danske Dagblades Forening [2012], ECLI:EU:C:2012:16; case C-360/13, Public
Relations Consultants Association Ltd v Newspaper Licensing Agency Ltd and Others [2014], ECLI:EU:C:2014:1195.
65 Triaille et al. 2014.
66 ECJ, C-302/10, Infopaq International A/S v Danske Dagblades Forening, para. 54.
67 Ibidem.
68 Triaille et al. 2014, pp. 31-32.
69 The “normal use of the content of the database has been established, for example, in the national
proceedings of the Ryanair case (ECJ, Case C-30/14, Ryanair Ltd v PR Aviation BV [2015],
ECLI:EU:C:2015:10). In that case, the Netherlands court stated that the activity of an online intermediary
CRIDES Working Paper Series 2018
reference to Art. 8 Database Directive, the TDM tool would likely extract and mine information
from an insubstantial part of the database (the “Legal conditions” section), without conflicting
with the normal exploitation of the database or unreasonably prejudicing the legitimate interests
of the maker or the author of the works or subject matter contained in the database. At the same
time, in some cases it will be difficult to allege this, as an online platform like Uber has 48
different legal documents (between T&C and a variety of “contractual” policies) that will not
qualify as insubstantial content.
Therefore, the Database Directive leaves some room to perform TDM, especially because
it expressly protects statutory permitted uses from contrary contractual provisions. However, this
balance of interests is ensured as long as there is a protection by the Database Directive. As the
Ryanair decision established, if the database is not protected either under copyright or the sui
generis right, the database owner can set down contractual limitations to its use.70
4. TDM exception and the draft Copyright in the DSM Directive: a new hope?
If the current system does not fit for the purpose of TDM, shall a specific exception
like the one contained in the currently debated proposal for a Directive on Copyright in the
DSM – be able to restore the balance between IPRs, on the one hand, and the access to
information and encouragement of AI innovation, on the other?
The current proposal arrives at the end of a process initiated by Commission with the
Communication for a Digital Single Market Strategy for Europe (2015)71, followed by the
Communication Towards a modern, more European copyright framework” (2015)72, and
confirmed in the Communication “Promoting a fair, efficient and competitive European
copyright-based economy in the Digital Single Market”.73 In those preparatory documents, the
comparing the prices of flights, by performing also the extraction of information from the Ryanair website,
was a “normal use” of that database, thus any contrary contractual provision considered unenforceable.
However, the ECJ noted that the existence of the sui generis right was not proven in the case: as a consequence,
the prohibition of contractual overriding did not apply.
70 ECJ, Case C-30/14, Ryanair Ltd v PR Aviation BV, paras. 44-45. Cf. Borghi and Karapapa 2015; Myska and
Harasta 2016.
71 Communication from the Commission to the European Parliament, the Council, the European Economic
and Social Committee and the Committee of the Regions, “A Digital Single Market Strategy for Europe”,
COM/2015/0192 final.
72 Communication from the Commission to the European Parliament, the Council, the European Economic
and Social Committee and the Committee of the Regions, Towards a modern, more European copyright
framework”, COM/2015/0626 final.
73 Communication from the Commission to the European Parliament, the Council, the European Economic
and Social Committee and the Committee of the Regions, “Promoting a fair, efficient and competitive
European copyright-based economy in the Digital Single Market”, COM/2016/592.
Limitations to text and data mining and consumer empowerment
focus of the Commission has always been on the need to promote innovation in research,
threatened by an uncertain legal framework for TDM and national differences.
In the September 2016 proposal for a Directive on Copyright for the Digital Single
Market (“Draft Copyright in the DSM Directive” or “Draft Directive”), the Commission
therefore introduced a specific exception for TDM. The exception expressly applies to the acts
of reproduction and extraction contemplated in Art. 2, InfoSoc Directive (right of reproduction),
Art. 5(a), Database Directive (temporary or permanent reproduction of the database), Art. 7(1),
Database Directive (extraction of the whole or a substantial part of it), Art. 11(1) Draft Directive
(right of reproduction and making available to the public, recognised to publishers of press
The proposal essentially allows research organisations to text and data mine works to
which they have lawful access for the purposes of scientific research. Research organisations are
defined as “a university, a research institute or any other organisation the primary goal of which
is to conduct scientific research or to conduct scientific research and provide educational
services: (a) on a non-for-profit basis or by reinvesting all the profits in its scientific research; or
(b)pursuant to a public interest mission recognised by a Member State; in such a way that the
access to the results generated by the scientific research cannot be enjoyed on a preferential basis
by an undertaking exercising a decisive influence upon such organisation”. Notably, Member
States should not provide for compensation for rightholders as regards uses under the TDM
exception (see Recital 13) and any contractual provision limiting TDM shall be unenforceable.
The norm is therefore both mandatory (for the Member States) and imperative (for private
parties), and there is no way to circumvent it through private ordering relying on contracts.
During the legislative procedure, the Council has specified the scope of the TDM and its
relationship with the proposal.74 First of all, by making clear that TDM is not always a copyright
relevant activity: “in relation to mere facts or data which are not protected by copyright […] no
authorisation is required under copyright law” (Recital 8a).75 Secondly, by reaffirming that acts
already covered by temporary reproduction exception under the InfoSoc Directive will continue
to benefit from that provision (Recital 8a).
Notably, the Council extends the TDM exception so as to include cultural heritage
institutions (“cultural heritage institution means a publicly accessible library or museum, an archive
74 See the version of the text dated 25 May 2018 (Council of the EU, Interinstitutional File: 2016/0280(COD),
doc. 9134/18, available here:
75 Here the Council tries to fix the wording of Recital 8 (Commission text). The latter has been criticised in
the literature for being a potential source of confusion, since it "wrongly suggests that carrying out TDM is per
se of relevance to copyright. The explanations given in Recital 8, according to which an authorisation to
undertake such acts must be obtained from rightholders if no exception or limitation applies, are too
sweeping”. Hilty and Richter 2017, p. 3. However, the clarification offered by the Council could not be
sufficient: the temporary reproduction exception is only one of the possible legitimate activities that can be
lawfully performed by users without authorisation nor a specific TDM exception.
CRIDES Working Paper Series 2018
or a film or audio heritage institution”).76 It added a security requirement in a new Art. 3(1a),
specifying also that the copies of the works and other subject-matter generated through TDM
shall not be retained for longer than necessary for achieving the purposes of scientific research
(see also Recital 11c).77
The text of the Council adds a further exception for TDM (new Art. 3a): Member States
are free to allow the “temporary reproduction and extraction of lawfully accessible works and
other subject-matter that form an integral part of the process of text and data mining”, if such
use “has not been expressly reserved” by the rightholder “including by technical means” (see also
Recital 13a). This seems to make the exception dependent on the rightholder’s willingness to
accept it, and the application of technical means would be enough to express it. Contrary than
the TDM exception provided in the initial text (Art. 3), it is an optional exception for the
Member States. This is not welcome as it will reinforce the risk of possible divergences between
Member States on an issue that should be dealt with seamlessly across the EU internal borders.
Furthermore, the additional TDM exception, if implemented, would be overridable (as Art. 6(1)
of the Commission’s draft has not been extended to cover the newly proposed Art. 3a). Anyway,
the possibility for rightholders to make some reservation risks to subordinate the legislative
exception to some private will. On the positive side, the scope of such additional exception is
much broader than the one proposed by the Commission as its beneficiaries go well beyond
research institutions and, eventually, cultural heritage institutions. Furthermore, it is true that
Art. 3a mirrors the wording and the content of the temporary reproduction exception at Art.
5(1) InfoSoc Directive (“temporary reproduction”, “form an integral part of the process”), but
with one important difference: the economic independent significance is out of the picture.
The discussion within the European Parliament (EP) showed the genuine concerns by
some EP members about the narrow scope of the TDM exception.78 However, many of the most
76 Defined at Art. 2(3), Draft Directive.
77 The wording of the Recital is far from being clear. On the one hand, by recognising the importance of peer
review and verification, the proposal seems to allow the retention of the copies made under the exception “in
certain cases” (not specified). Such copies must be stored in a secure environment and not be retained for
longer than is necessary for the scientific research activities”. The text leaves to the Member State the task to
determine the concrete modalities for retaining the copies. However, the hard issue to determine is: when will
the copies no longer be necessary? It is interesting to note a parallelism: the GDPR has ensured through several
provisions that data should not be stored longer than necessary with respect to the purpose of the processing.
However, scientific research is one of the cases that justifies the possibility to make an exception to that rule
(see, Art. 5(1)(e), GDPR). Interestingly, the scientific research can trump storage limitations when the data
subjects’ interests are at stake, but not in the IP domain when data aggregators prerogatives are involved.
78 See, for instance, amendments proposed to the text of Art. 3 of the Commission proposal: amendment 538
by Julia Reda, Nessa Childers, Max Andersson, Michel Reimon, Brando Benifei (deleting any reference to
research organisation, research purposes and lawful access), amendment 539 by Jytte Guteland (extending
TDM to cultural heritage institutions), amendments 546 and 547 (respectively encouraging and obliging
Members States to allow research organisations, without lawful access to works and other subject-matter, to
perform TDM), amendment 548 (protecting the mandatory TDM exception against TPMs) and amendments
551-555 (limiting the scope of the measures that the rightholder can adopt to ensure the security and integrity
of the networks and databases where the works or other subject-matter are hosted), amendment 564
Limitations to text and data mining and consumer empowerment
favourable amendments to TDM were not tabled in the version the Committee on Legal Affairs
(JURI) presented to the EP plenary for the vote of July, 5.79
The JURI report reflected the dual approach followed by the Council in the text adopted
in May 2018, proposing a mandatory TDM exception in favour of research organisations for the
purpose of scientific research, provided that they have lawful access to the works or other subject-
matter (Art. 3), and an optional exception available to anyone as long as the rightholder has not
reserved the use of works and other subject-matter in a machine readable format (Art. 3a). Query
what those reservations in a machine readable format are and how they could be implemented.
The main reservation we have with this part of the EP resolution on the copyright reform is that
the TDM exceptions could still be overridden by contracts (Art. 3a) and/or technical means
(Arts. 3 and 3a).
Furthermore, the JURI version added two twin provisions at both Art. 3 and Art. 3(a),
allowing Member States to introduce respectively mandatory or optional TDM exceptions in
accordance with Art. 5(3)(a) InfoSoc Directive, which refers to the teaching and scientific
research exception for original works and other subject-matter. These provisions have been
probably designed for preserving the TDM exceptions already adopted by some Members States
(for instance, UK, Estonia, France, Germany). However, such legislative choice is likely to
fragment the European legal framework for TDM, allowing the blossoming of diverse national
solutions and increasing the divide between TDM acts over works and databases (since the
possibility refers uniquely to the research exception under the InfoSoc Directive).
Finally, with reference to the boundaries of the copyright relevant activity in TDM
processes, the EP version introduced a new specific provision at Recital 8a. Dealing with the
technical aspects of TDM, such amendment states that TDM as mere reading and analysis of
digitally stored, normalised information” is not a copyright relevant act. Copyright may come
into play only in case of reproductions or extractions linked to the access and process of
“information normalisation”. The latter is not expressly defined, but as it emerges from the same
recital, it refers to the preparatory activities which enable the automated computational analysis,
such as the change of the format of information or the extraction from a database into another
one that will be subjected to TDM. The exceptions provided in the Draft Directive, therefore,
will cover only these activities. Apparently, this recital seems to restrict the scope of the exception.
On the contrary, it reaffirms that not all TDM’s stages need a specific exception to be valid: many
(mandating the adoption of open formats for publicly-funded research and data in order to enable TDM). The
text of the amendments presented by the Member of the EP is available here:
79 Committee on Legal Affairs, Report on the proposal for a directive of the European Parliament and of the
Council on copyright in the Digital Single Market (COM(2016)0593 C8-0383/2016 2016/0280(COD)),
29 June 2018, available here:
CRIDES Working Paper Series 2018
of them (such as the analysis, the creation of patterns and the subsequent publication) can be
done freely and without infringing any IP entitlement.
The European Parliament voted against the JURI version during the plenary of July 2018,
postponing any further decision to the September’s meeting. However, in September, the EP
approved the TDM amendments that essentially reproduce the July’s version.80 The Copyright
reform is now in the trilogue’s phase of discussions between the Commission, the Council and
the EP.
It thus appears that the various texts on TDM available in September 2018 provide for
the following exceptions:
Commission text
Council text
EP text
TDM for research
Mandatory and not
Mandatory and not
Mandatory and not
Member States may
continue to provide
TDM exceptions in
accordance with Art.
5(3)(a) of InfoSoc
TDM for other
(including in a
commercial context)
Optional and
Optional and
Member States may
continue to provide
TDM exceptions in
accordance with Art.
5(3)(a) of InfoSoc
The Commission, the Council and the Parliament texts are unsatisfactory to promote
the development of Big Data analytics in Europe. The conditions on the objectives of TDM (not-
for-profit research or research with a public interest goal) and the beneficiaries (research
organisations and cultural heritage institutions) lead to a narrow exception.81 First, because TDM
80 Amendments adopted by the European Parliament on 12 September 2018 on the proposal for a directive
of the European Parliament and of the Council on copyright in the Digital Single Market (COM(2016)0593
C8-0383/2016 2016/0280(COD)), available here:
81 In this sense, see the second sentence added at para. 1 of Art. 3, EP text.
Limitations to text and data mining and consumer empowerment
(and research in general!) is not exclusive of the academic circles.82 TDM could be used by policy
makers to test draft policies and regulation, by journalists or private individuals for fact checking,
by consumers and lawyers to automatically compare the terms of service of different platforms,
just to give a few examples. Such entities and individuals will not be able to acquire additional
knowledge and/or provide additional services and tools relying on the TDM insights.
Second, in many instances (even in academic research which is partly supported by
private funding), the boundary between commercial and non-commercial research is not always
easy to trace. Such a limitation is likely to undermine also the spirit of the initial goal of the
Commission, anticipated in the Communication for a Digital Single Market Strategy, whose intention
was to promote research innovation for both non-commercial and commercial purposes.83
Third, even if contractual limitations are pushed out of the door, private ordering can
come back through the window, via TPMs. The proposal, in fact, does not grant any effective
protection against TPMs as it is not clear there is a possibility to legally circumvent those that
would unlawfully limit TDM.84
Finally, even if the copyright reform will expand TDM boundaries, the main obstacle
remains the ECJ Ryanair holding and the power of contracts in the absence of copyright or
database protection. Of course, it does not make much sense to address this issue in an
instrument, as the Draft Directive, dealing with intellectual property as the problem is to secure
some uses when there is no IP protection but such issue risks never being addressed in another
context either.
82 Many scholars have argued that the exception should be broadened. For instance, Margoni and Kretschmer
2018; Caspers and Guibault 2016b; Margoni and Dore 2016; Hilty and Richter 2017; Geiger et al. 2018.
83 See above Communication, “A Digital Single Market Strategy for Europe”, COM/2015/0192 final, p. 7:
Innovation in research for both non-commercial and commercial purposes, based on the use of text and data
mining (e.g. copying of text and datasets in search of significant correlations or occurrences) may be hampered
because of an unclear legal framework and divergent approaches at national level. The need for greater legal
certainty to enable researchers and educational institutions to make wider use of copyright-protected material,
including across borders, so that they can benefit from the potential of these technologies and from cross-
border collaboration will be assessed, as with all parts of the copyright proposals in the light of its impact on
all interested parties”.
84 Margoni and Kretschmer 2018; Caspers and Guibault 2016a. As acutely pointed out by Margoni and
Kretschmer: “The EU legislature is fully aware of this contradiction but failed to address it properly. In fact,
Art. 6 of the Proposal (“common provisions”) clarifies that the provisions of the first, third and fifth
subparagraph of Art. 6(4) InfoSoc directive apply. In plain English, this means that if a user qualifies for an
exception to copyright (e.g. TDM) but a Technological Protection Measure prevents them from doing it,
Member States have an obligation to take appropriate measures to ensure that right holders make available to
the beneficiary an exception or limitation. In the almost 20 years since when the InfoSoc directive was enacted,
the UKIPO, which has correctly put in place a specific procedure for this type of situations, has received less
than a handful of requests”.
CRIDES Working Paper Series 2018
5. Reconstructing the reproduction right: back to the roots
The TDM exception, if it is well designed at the end of the legislative process, will
somewhat improve the situation but it is not likely to be the panacea. Some issues, like the one
of SDSs that prompted our paper, will remain if the provision is overridable and not broad. It is
probably too little considering the downside of the exercise: by searching for some legal certainty,
guarded by a statutory exception, the EU legislator will confirm in black letter that TDM, even
with the clarifications made by the EP in Recital 8a, is a copyright-relevant activity. The risk we
are running is to give up on a crucial point, instead of embracing the fatigue of a serious
discussion about the boundaries of the right of reproduction and copyright foundations.85
That copyright might extend to the TDM process does not appear legitimate: no author
of literary or artistic production in the past has conceived her copyright as a way to limit the use
of her work as a source of useful information, for instance for discerning fluctuations in interest
in a particular subject or for determining fashionable expressions. No author, while producing a
work, has seriously relied on the possibility of earning revenues from the derivative use related
to searching and indexing a corpus that includes her work.86 Such use is far removed from the
core exploitation field of most works. In addition, when it appeared more than two centuries
ago, copyright was not only intended to remunerate authors (and publishers) for creating (and
disseminating) works, but was also promoted to expand public learning. This is not only true for
the US copyright system, as the same rationale is at the origin and core of the continental droit
d’auteur system.87
Even without an express exception, we can rely on the implicit requirement that the
reproduction involves a use as a work. Such use as a work does not exist in the case of TDM, nor
in other cases involving copying for deriving information or checking conduct (e.g., to identify
plagiarism). As put by the Court in Authors Guild v. Google, Inc., ‘the purpose of Google’s copying
of the original copyrighted books is to make available significant information about those books,
permitting a searcher to identify those that contain a word or term of interest’ (emphasis
added).88 This purposive analysis of the act of copying strongly weighs in favour of fair, because
highly transformative, use. In the EU, the requirement of use as a work can help to reach the
same outcome. Indeed, when acts of reproduction are carried out for the purpose of search and
TDM, the work is not used as a work, it only serves as a tool or data for deriving other relevant
information. The expressive features of the work are not used, and there is no public to enjoy
85 These thoughts were firstly elaborated in Strowel 2018.
86 See also Poort 2018.
87 See the review of the history and principles of copyright in both legal traditions: Strowel 1993. More recently,
Strowel 2014, pp. 701703.
88 Authors Guild v. Google, Inc. No. 13-4829-cv (2d Cir. Oct. 16, 2015). On April 18, 2016, the Supreme Court
denied the petition for a writ of certiorari, leaving the Second Circuit ruling in Google's favour intact. To
make available parts of the corpus of books, Google has scanned the digital copies and established a publicly
available search function, the ngrams tool.
Limitations to text and data mining and consumer empowerment
the work, as the work is only an input in a process for searching a corpus and identifying
occurrences and possible trends or patterns.
Even if the EU Parliament and Council decide to include an exception for TDM that is
broad enough (this would require some commercial research to be exempted), the analysis based
on ‘the use of the work as a work’ condition for copyright infringement remains necessary to
address other types of copying for the purpose of providing information, such as copies for
checking mistakes and plagiarism, copies to use or repair a protected work with a utilitarian
function, non-transitory copies made on proxy servers, smart disclosure systems and many other
uses that cannot be anticipated.89
As said, the main reason behind the introduction of the TDM exception was the
reduction of legal uncertainties and the diverging national implementations regarding the
research exceptions. If we look at the texts to be discussed during the trilogue, such uncertainties
remain, and more divergences are likely because of the optional nature of the TDM exception
provided in favour of beneficiaries other than research organisations.
6. Prohibition of TDM in T&C
As already mentioned, contracts also can affect TDM. In Sections 3 and 4, we have pointed
out that, in some cases, copyright exceptions cannot be limited via contractual means. However,
the issue is not harmonised under the InfoSoc Directive and even when the exceptions are
protected against the power of contract, the free room they provide might be annihilated by
TPMs. Furthermore, according to the Ryanair ruling, the copyright (or IP) balance cannot be
exported so as to apply when there is no copyright (IP), but only contracts between parties.
To give an idea of the scope of the issue, we analyse in this Section the occurrences of
contractual clauses prohibiting TDM or otherwise restricting copyright exceptions. As explained
in Section 3, we looked at the copyright and database protection provisions, since TDM usually
involves a “reproduction” and/or an “extraction” of the subject-matter to mine.90
89 The following US decisions quoted in Authors Guild v. Google, Inc. have, for instance, exempted several
uses that, without the application of the work use requirement, cannot escape copyright’s exclusivity in the
EU: A.V. ex rel. Vanderhye v. iParadigms, LLC, 562 F.3d 630, 638640 (4th Cir. 2009) (justifying as
transformative fair use purpose the complete digital copying of a manuscript to determine whether the original
included matter plagiarized from other works); Perfect 10, Inc. v., Inc., 508 F.3d 1146, 1165 (9th
Cir. 2007) (justifying as transformative fair use purpose the use of a digital, thumbnail copy of the original to
provide an Internet pathway to the original); Kelly v. Arriba Soft Corp., 336 F.3d 811, 818819 (9th Cir. 2003)
(same); Bond v. Blum, 317 F.3d 385 (4th Cir. 2003) (justifying as fair use purpose the copying of author’s
original unpublished autobiographical manuscript for the purpose of showing that he murdered his father and
was an unfit custodian of his children).
90 Caspers and Guibault 2016a; Stamatoudi 2016; Triaille et al. 2014.
CRIDES Working Paper Series 2018
The analysis focused on the terms and conditions (T&C) of twenty-one online platforms,
equally distributed among three sectors: mobility (carpooling and car sharing), accommodation
(including services for sharing office space), food (i.e., initiatives for the recuperation of unsold
or unused food, sharing or delivery of home-cooked meals, etc.). In the selected sample, we tried
to ensure a good balance between platforms operating globally and locally and between large
capitalistic and cooperative initiatives.91 If different contractual versions were available, we have
consulted the T&C for the Belgian market.
As shown in Table 1, 20 out of 21 platforms published the T&C on their website and
14 of them contained specific intellectual property clauses, directly or indirectly, related to TDM
activities. More specifically:
- four platforms expressly prohibit TDM on the website content92;
- Three others do not allow the use of any kind of bot, crawler or scraper (i.e., the
automated software agents that search through the content of webpages. These are
necessary tools for TDM, e.g. when there is no available application programming
- in four cases, the reproduction or copy of website materials which is usually a
preliminary step of TDM process – is forbidden; and
- in three occasions, the formulation was vague or broad enough to exclude TDM.94
This shows a trend toward a general contractual ban of TDM. The prohibition is broad
and refers to all the website’s contents and services, thus including the informative pages
containing the legal conditions. Such provisions were inserted with a view to protect the data (in
some ways valuable) related to the service (e.g. the timetable of flights, the prices, the list of
accommodation and contact details of the owners, etc.) from free exploitation: their effect on
T&C and privacy policy would be an involuntary side effect. However, this is not the case: the
contractual provision is usually confirmed and embedded into the technical instructions of the
To check this, we used the robots.txt file. It is an exclusion protocol that content
providers can insert into the root directory to prevent crawling or indexing activities on certain
91 On the theoretical foundation of platform cooperativism, see Scholz 2016.
92 Blablacar does not prohibit TDM on the whole content of the website, but only on a substantial part of it.
93 On crawling and scraping see Caspers and Guibault 2016a, p. 8.
94 According to T&C of Bar d’Office, users cannot obtain (or attempt) to obtain any material or information
through any means not intentionally made available by the platform. We shall conclude that third party
applications aiming at analysing Bar d’office legal documents do not fall within the permitted uses. Wibee
does not allow to “exploit in any way the content” but it has to be combined with the contractual provision
that allows the use of the platform for non-commercial purposes only. Finally, Menu Next Door contained the
broad, but vague, formulation “All rights reserved”.
Limitations to text and data mining and consumer empowerment
pages of their website.95 Adding the extension “robots.txt” at the end of the address of the website
is possible to see the underlying instructions.96 The latter can consist of two main commands:
- User-agent: it shows at what robots the instructions are directed. If there is an asterisk
(“User-agent: *”), this means that the section applies to all robots.
- Disallow: it indicates what pages cannot be visited by the robots.
We found that three platforms are actually using the exclusion protocol to keep robots
away from the whole server, one from a directory which contains amongst others the legal
documents, and two specifically from the page of T&C and privacy policy (see the annexed Table
1, “Robots.txt” column). In all cases where the robots were disallowed, the T&C provisions
prohibited TDM as well. Only in one case there was no TDM-related provision in the contract,
but nevertheless the code instructions did not allow the indexing of the website content by a
series of robots. However, it should be noticed that the robots excluded are those usually
mentioned in the “black lists” of malicious programs.
The nature of robots.txt as a TPM is controversial. Looking at the definition of
“technological measures” provided at Art. 6 of the InfoSoc Directive, the robots.txt could fall
under it. Indeed, the protocol is a kind of technology that “is designed to prevent or restrict acts,
in respect of works or other subject-matter, which are not authorised by the rightholder of any
copyright or any right related to copyright”97. However, the effectiveness of such technological
measures is debatable. According to the Directive, it implies that “the use of a protected work or
other subject-matter is controlled by the rightholders through application of an access control or
protection process, such as encryption, scrambling or other transformation of the work or other
subject-matter or a copy control mechanism, which achieves the protection objective”.98 Some
authors have argued that the protocol contains instructions that do not qualify as a technical
barrier: any software agent can simply ignore the “Disallow” command without actively forcing
any digital fence.99 The content provider has just to rely on the voluntary compliance of the user,
hoping that the visiting agent has been designed to follow the ASCII syntax. In any case, even if
the robots.txt is not considered as an effective technological measure or not, its use by the content
providers and online platforms confirms their willingness to limit TDM, as stated in their
95 On the origin and functioning of this file, see
96 Rotenberg and Compañó 2009.
97 Art. 6(3), InfoSoc Directive.
98 Ibidem.
99 Sire 2015. Contra, Groom 2004, according to which a visiting agent programmed to systematically ignore
the robots.txt can be seen as a strategy to circumvent a technological protection measure.
100 In Europe, the robots.txt file has been questioned with reference to the issue of implied license only. See,
for instance, the Copiepresse v. Google saga, commented in Strowel 2007, 2011. Doubts about the classification
of robot.txt as a TPM in the US context are expressed by Jasiewicz 2012. In the US case Healthcare Advocates,
Inc. v. Harding (497 F. Supp. 2d 627, 643 (E.D. Pa. 2007)), the Court for the Eastern District of Pennsylvania
CRIDES Working Paper Series 2018
How legitimate is such a general and absolute ban imposed in many T&C? The
prohibition of TDM is likely to undermine the legitimate activity of consumers that could use
smart disclosures mechanisms and instruments for the automatic analysis of contractual
documents to better understand the terms of the user agreement. If a sort of “bionic eye” is
available in order to scan a document and extract the relevant pre-contractual information, is it
justified to prohibit its use?
As shown in the previous Sections, neither the current framework of copyright
exceptions nor the new TDM exceptions under consideration by the Council and EP could
apply to SDSs and be able to protect the rightful interests of the users. However, the principle of
transparency, embedded in consumer and data protection legislations, could offer a last line of
defence against an absolute prohibition of TDM.
7. Transparency 2.0: a right to machine legibility
Transparency is a cardinal principle in EU Law.101 In consumer and data protection
legislations, information transparency is designed as a necessary feature of mandated disclosures,
i.e. the obligation to provide one party (traditionally the weak one) with the information
concerning the transaction. If the information is accessible, clear and understandable, mandated
disclosures can effectively inform the consumer or the data subject about the essential content
of the agreement and allow her to make an optimal decision or express a meaningful consent,
where requested. In this sense, the substantive requirement (the duty to provide certain
information) is complemented by formal requirements (the provision of the information in
advance and the use of plain and intelligible language).
In consumer law, the principle of transparency has been traditionally interpreted as
comprising two components: 102 1) the consumer has to be able to have knowledge of the terms
before entering into contract; 103 2) the information has to be provided in a way that the average
incidentally discussed the nature of the robots.txt file. The judge recognised the protocol as a TPM under the
DMCA in that specific case. However, the Court expressly affirmed that robots.txt is not “analogous to digital
password protection or encryption” and its nature must be assessed case-by-case (“This finding should not be
interpreted as a finding that a robots.txt file universally qualifies as a technological measure that controls access
to copyrighted works under the DMCA”).
101 Buijze 2013. The analysis presented in this paragraph has been further developed in Ducato forthcoming.
102 Loos 2015; Kästle-Lamparter 2018, pp. 429-430, 474 and 481. See also, Micklitz et al. 2009, pp. 135 ff.
103 The transparency principle, as a duty to provide information before the conclusion of the contract, is
envisaged in the Annex of the Unfair Terms Directive (UTD), which includes, among the list of potential
unfair terms, the contractual provision which: “irrevocably binds the consumer to terms with which he had
no real opportunity of becoming acquainted before the conclusion of the contract” (Annex, 1.i, UTD). It can
also be derived by Recital 20, UTD. It is further recalled at Art. 6(1) of the Consumer Rights Directive (CRD).
Limitations to text and data mining and consumer empowerment
consumer can understand without a legal advice. The latter means that information must be
legible and given in a plain and intelligible language.104
The principle of legibility, in particular, requires to take into consideration the font size,
the layout and the accessibility of pre-contractual information. As pointed out by Micklitz et al.,
the possibility to actually read the text of the contract, i.e. the design of conditions “plainly both
from an editing and optical point of view”105 (no “small-print” for instance), must be seen as a
corollary of intelligibility enshrined in the 93/13/EEC Unfair Terms in consumer contracts
Directive (“UTD”).
The principle of legibility has been expressly codified in the 2011/83/EU Consumer
Rights Directive (“CRD”) at Art. 7(1) for off-premises contracts and Art. 8 (1) for distance
contracts, which both state that information shall be legible if provided on a durable medium.
Interestingly, Kästle-Lamparter points out that: “Taken at a face value, the requirement of
legibility excludes mere audio tapes or files: information must be provided in a human-readable
format. But perhaps this was not intended and the provision should be rather read ‘legible if
text-based’, or possibly ‘legible or audible’ ”.106 Furthermore, it should be noted that if we assume
the additional application of Directive 2000/31/EC (“e-commerce Directive”) when the contract
is concluded with an online platform, Art. 10(3) establishes that: “Contract terms and general
conditions provided to the recipient must be made available in a way that allows him to store
and reproduce them” [emphasis added]. The latter could therefore constitute an additional
argument to support the mining of T&C’s text.
The rationale of consumer protection, the ECJ’s interpretation of the principle of
transparency but also the formal requirements set out in the e-commerce Directive are likely to
104 The principle of transparency, sub specie of understandability of the information provided to the consumer,
is specifically mentioned in several legislative instruments. The duty to provide the consumer with information
in a clear and comprehensible manner is recalled at Art. 5, UTD: “In the case of contracts where all or certain
terms offered to the consumer are in writing, these terms must always be drafted in plain, intelligible language”.
Moreover, it is expressed at Arts 5(1), 6(1) CRD. and further expanded at Art. 8 CRD. In addition, when the
contract is concluded through a means of distance communication which allows limited space or time to
display the information” (Art. 8.4, CRD), like the screen of a mobile phone, the trader will have to provide at
least a set of pre-contractual information, such as the main characteristics of the goods or services, the identity
of the trader, the total price, the right of withdrawal, the duration of the contract and, if the contract is of
indeterminate duration, the conditions for terminating the contract. Among the appropriate means to display
information to the consumer, the Commission suggested the adoption of a set of icons, making also available
a model. However, such a measure does not seem to have taken hold (EC Commission, DG Justice Guidance
document concerning Directive 2011/83/EU of the European Parliament and of the Council of 25 October
2011 on consumer rights, amending Council Directive 93/13/EEC and Directive 1999/44/EC of the
European Parliament and of the Council and repealing Council Directive 85/577/EEC and Directive
97/7/EC of the European Parliament and of the Council, June 2014, available here:
105 Micklitz et al. 2009, p. 136.
106 Kästle-Lamparter 2018, p. 474.
CRIDES Working Paper Series 2018
accommodate a broad understanding of legibility.107 In particular, to ensure the balance of
interests embedded in the 1993 UTD and 2011 CRD, it will be coherent to frame the legibility
requirements, adopting a technologically neutral approach. This functional interpretation is
further confirmed in the GDPR.
The analogy with the transparency in the data protection framework, despite the distinct
area of application and the slightly different terminology used, is justified as there is a koinè, if
not an open dialogue, between these two legal branches.108
The GDPR has recently codified transparency as a pillar of data protection along the
principles of lawfulness and fairness of the processing (Art. 5.1.a, GDPR). As specified by the
European Data Protection Board (EDPB): “Transparency is an overarching obligation under the
GDPR applying to three central areas: (1) the provision of information to data subjects related
to fair processing; (2) how data controllers communicate with data subjects in relation to their
rights under the GDPR; and (3) how data controllers facilitate the exercise by data subjects of
their rights”109. The principle of transparency echoes in many aspects the articulation shown in
the consumer legislation.
The GDPR explicitly refers to legibility when it allows the use of standardised icons to
complement the privacy notice. Icons shall give “in an easily visible, intelligible and clearly legible
manner a meaningful overview of the intended processing”110. The GDPR further specifies that
“where the icons are presented electronically they shall be machine-readable” (Art. 12.7, GDPR).
107 When it has interpreted the plainness and intelligibility requirements, the ECJ has always excluded
formalistic readings. In Kásler, for instance, the Court hold that the requirement of transparency of terms,
under the UTD, cannot “be reduced merely to being formally and grammatically intelligible” (ECJ, C-26/13,
Árpád Kásler and Hajnalka Káslerné Rábai v OTP Jelzálogbank Zrt. [2014] ECLI:EU:C:2014:282, para. 71).
Principle confirmed in the subsequent jurisprudence. See, ECJ, Bogdan Matei and Ioana Ofelia Matei v SC
Volksbank România SA [2015] ECLI:EU:C:2015:127. See also, ECJ, C-191/15, Verein für
Konsumenteninformation v Amazon EU [2016] ECLI:EU:C: 2016:612). Terms must be transparent “so that
the consumer can foresee, on the basis of clear, intelligible criteria, the economic consequences for him which
derive from it” (ECJ, C-26/13, Árpád Kásler and Hajnalka Káslerné Rábai v OTP Jelzálogbank Zrt, para. 73). As
later specified in Gutiérrez Naranjo, under the transparency principle, the consumer has to be able to
understand not only the economic consequences but also the legal ones (ECJ, C-154/15, Francisco Gutiérrez
Naranjo v Cajasur Banco SAU, Ana María Palacios Martínez v Banco Bilbao Vizcaya Argentaria SA (BBVA), Banco
Popular Español SA v Emilio Irles López and Teresa Torres Andreu [2016] ECLI:EU:C:2016:980).
108 About the interplay between consumer protection and data protection, Helberger et al. 2017.
109 EDPB, Guidelines on Transparency under Regulation 2016/679, 13 April 2018.
110 Art. 12.7 GDPR. See also Recital 60. The link between transparency, information and visualisation is
further stressed at Recital 58, GDPR: “The principle of transparency requires that any information addressed
to the public or to the data subject be concise, easily accessible and easy to understand, and that clear and
plain language and, additionally, where appropriate, visualisation be used. Such information could be provided
in electronic form, for example, when addressed to the public, through a website. This is of particular relevance
in situations where the proliferation of actors and the technological complexity of practice make it difficult for
the data subject to know and understand whether, by whom and for what purpose personal data relating to
him or her are being collected, such as in the case of online advertising”.
Limitations to text and data mining and consumer empowerment
This is a highly behaviourally-informed legal innovation, considering that more often online
users derive information from icons and pictures.111
However, if the GDPR is systematically interpreted, the principle of transparency emerges
in several other loci. First of all, data controllers, i.e. the person or entity which determines the
purposes and means of the processing of personal data, shall take appropriate measures to
provide the information required by law (at Arts. 13 and 14 GDPR) and any communication
regarding the right of access, the use of automated individual decision-making, and personal data
a) in a concise, transparent, intelligible and easily accessible form;
b) using clear and plain language;
c) provided in writing or by other means, including, where appropriate, by electronic means;
d) provided orally, if requested so by the data subject.
Furthermore, where the processing is based on the data subject’s consent, the request for
it has to be presented in “a manner which is clearly distinguishable from the other matters, in
an intelligible and easily accessible form, using clear and plain language”113.
The Guidelines on Transparency (2018) drafted by the EDPB offer a first reading of such
requirements. If according to “concision and transparency”, information has to be presented
“efficiently and succinctly, in order to avoid information fatigue”114, the “easy accessibility”
imposes to make information clearly visible, e.g. on the website. This means that data controllers
have to actively furnish the information (or the way to find it), while it is not an obligation of
the data subject to start a quest for retrieving information. For example, data controllers should
provide information by giving it directly to data subjects, “by linking them to it, by clearly
signposting it or as an answer to a natural language question”115. In addition, the duty to provide
information in “writing or by other means” specifically recalls the CRD obligation to make
information available to the consumer “in a way appropriate to the means of distance
111 OECD 2016, p. 17. A version for the GDPR set of icons was released in the Annex of the first reading of
the European Parliament (European Parliament legislative resolution of 12 March 2014 on the proposal for a
regulation of the European Parliament and of the Council on the protection of individuals with regard to the
processing of personal data and on the free movement of such data (General Data Protection Regulation)
(COM(2012)0011C7-0025/2012 2012/0011(COD)) (Ordinary legislative procedure: first reading). But no
other official initiatives have been taken in this direction. An interesting methodology to answer the challenges
of GDPR’s icons has been developed within the research project run by the Cirsfid group at the University of
112 Art. 12.1, GDPR. See also Recitals 39 and 42, GDPR.
113 Art. 7.2, GDPR.
114 EDPB, Guidelines on Transparency under Regulation 2016/679, p. 8.
115 Ibid., p. 11.
CRIDES Working Paper Series 2018
communication used” (Art. 8.1, CRD). The EDPB emphasises that such a provision has to be
interpreted broadly, allowing to choose the most appropriate means and format to reach the
informative goal: for a privacy policy on websites, “digital layered privacy, but also ‘just in time
contextual pop-up notices, 3D touch or hover-over notices, and privacy dashboards”116 can be
The Board further underlines the importance to consider the specific circumstances
where the provision/communication occurs or how the interactions between the controller and
the data subject happen. For instance, the EDPB warns that the electronic privacy notice on a
website is likely to be ineffective for screenless IoT or smart devices: it could be preferable, for
example, to include the privacy policy in the instruction manual or make it easier to access
through a QR code printed on the device.117
To sum up, through the transparency principle, sub specie of obligation to provide concise
and transparent, easy accessible information, in writing or other means, the GDPR recalls the
conceptual background already seen in consumer law. Furthermore, the EDPB encourages the
use of technological advancements to inform users in a more meaningful way.
If we look at these provisions systematically and at their interpretation by EDPB and the
ECJ, they can operate as functional equivalents: both of them pursue the same rationale, i.e.,
protecting the weak party from information asymmetries, by requiring that information must be
visible, accessible and readable.
Therefore, if this is the policy goal, the principle of transparency enshrined in consumer
and data protection laws is technologically neutral and can accommodate a right to “machine
legibility”. By machine legibility we mean the possibility for a SDS to have access to the pre-
contractual information (T&C) and the information related to the processing (privacy policy) in
a format processable by the smart system. If thanks to AI there are now instruments enhancing
the human ability to read and understand contractual terms and privacy settings, not only
information should be legible to a human eye, but also to the tool that a user can take advantage
of. Therefore, either depending on IPRs, contractual or TPM limitations, an absolute
prohibition of TDM on pre-contractual and privacy information available online on the website
of the platform will unreasonably restrict a legitimate prerogative of the consumer or the data
Furthermore, the solution will also be costless to the data controller/trader: the platform
could allow TDM, by enabling the indexing of the corresponding page on their website through
the robots protocol and lowering TPMs barriers, if any.
116 Ibid., p. 12.
117 Ibidem.
Limitations to text and data mining and consumer empowerment
8. Conclusion
SDSs offer clear benefits for the consumers as they reduce information asymmetries and
improve the access to data needed to take informed decisions. The possibility to deploy SDSs
might however be prevented as there are technical and legal obstacles, including the prohibitions
inserted in many T&C and other online documents of interest for the consumers. In this
context, it is not clear whether TDM tools needed for the use of SDSs can be used. The existing
copyright and database exceptions do not adequately tackle the TDM issue, and, as recognized
by the European Commission, there is a need for introducing a TDM exception. However,
neither the initial proposal by the European Commission focusing on the research context, nor
the amendments discussed within the Council and the European Parliament appear sufficient
to facilitate the use of TDM for improved smart disclosure and, more broadly, for AI
Regarding SDSs, the transparency principle embedded in consumer and data protection
rules offers some legal ground to justify the use of TDM for consumer empowerment. But the
requirement of machine legibility that appears necessary in a society where the automatic
treatment of information becomes central could be further promoted by a well-designed TDM
exception. We are not yet there.
CRIDES Working Paper Series 2018
Ayres I, Schwartz A (2014) The No-Reading Problem in Consumer Contract Law. Stanford Law
Review, 66(3): 545-610
Bakos Y, et al. (2014) Does Anyone Read the Fine Print? Consumer A ention to Standard Form
Contracts. The Journal of Legal Studies 43(1): 1-35
Bar-Gill O (2015) Defending (Smart) Disclosure: A Comment on More Than You Wanted to
Know. Jerusalem Review of Legal Studies 11(1): 75-82
Ben-Shahar O (2009) The Myth of the ‘Opportunity to Read’ in Contract Law. European Review
of Contract Law 5(1): 1-28
Ben-Shahar O (2013) Regulation Through Boilerplate: an Apologia. Mich. L. Rev. 112: 883-903
Bently L, Sherman B (2009) Intellectual property law. Oxford University Press, Oxford
Bernault C, et al. (2017) Traité de la propriété littéraire et artistique. Litec, Paris
Bernhardt B, et al. (2015) Revolutionizing Scholarship: a Panel Discussion on Text and Data
Mining. Serials Review, 41(3): 184-186
Beunen AC (2007) Protection for databases: The European database directive and its effects in
the Netherlands, France and the United Kingdom. Wolf Legal Publishers, Nijmegen
Borghi M, Karapapa S (2015) Contractual restrictions on lawful use of information: sole-source
databases protected by the back door?. European Intellectual Property Review 37(8): 505-
Braun D, et al. (2018) Customer-centered LegalTech: Automated Analysis of Standard Form.
Internationales Rechtsinformatik Symposium (IRIS): 627-634
Buijze A (2013) The six faces of transparency. Utrecht L. Rev. 9(3): 3-25
Busch C (2016) The future of pre-contractual information duties: from behavioural insights to
big data. In: Twigg-Flesner C (ed), Research Handbook on EU Consumer and Contract
Law. Edward Elgar, Cheltenham, UK, Northampton, MA, USA, 221-240
Busch C (forthcoming) Implementing Personalized Law: Personalized Disclosures in Consumer
Law and Privacy Law.
Accessed 18 August 2018
Caso R (2004) Digital rights management: il commercio delle informazioni digitali tra contratto
e diritto d'autore. CEDAM, Padua
Limitations to text and data mining and consumer empowerment
Caspers M, Guibault L (2016a) Baseline report of policies and barriers of TDM in Europe.
Policies-and-Barriers-of-TDM-in-Europe-1.pdf. Accessed 18 August 2018
Caspers M, Guibault L (2016b). A right to ‘read’ for machines: Assessing a black-box analysis
exception for data mining. Computer Sciences 53(1): 1-5
Contissa G, et al. (2018) CLAUDETTE meets GDPR. Automating the Evaluation of Privacy
Policies using Artificial Intelligence.
066_claudette_meets_gdpr_report.pdf. Accessed 18 October 2018
de Visscher F, Michaux B (2000). Précis du droit d'auteur et des droits voisins. Bruylant, Brussels
Derclaye E (2008) The Legal Protection of Databases: A Comparative Analysis. Edward Elgar,
Cheltenham, UK, Northampton, MA, USA
Derclaye E (2012). Football Dataco: skill and labour is dead.
dead/. Accessed 18 August 2018
Derclaye E (2014) The Database Directive. In: Stamatoudi I, Torremans P (eds.), EU copyright
law: A commentary. Edward Elgar, Cheltenham, UK, Northampton, MA, USA, 298-354
Derclaye E, Favale M (2010) Paper 3: User Contracts. In: Kretschmer M, et al. (eds.) The
relationship between copyright and contract law. Accessed 18
August 2018
Ducato R (forthcoming) Transparency by (Legal) Design. Paper presented at the Private Law
Consortium, Harvard Law School, 15 May 2018 and at the Younger Scholar Informal
Symposium, organised within the General Congress of the International Academy of
Comparative Law, Fukuoka, 25 July 2018
Dussolier S (2007) Sharing Access to Intellectual Property through Private Ordering. Chicago-
Kent Law Review 82(3):1391-1438
Feldman R, Sanger J (2007). The text mining handbook: advanced approaches in analyzing
unstructured data. Cambridge University Press, Cambridge
Geiger C, et al. (2018) The Exception for Text and Data Mining (TDM) in the Proposed
Directive on Copyright in the Digital Single Market - Legal Aspects. Centre for
International Intellectual Property Studies (CEIPI) Research Paper No. 2018-02: 1-34
Groom J (2004) Are Agent Exclusion Clauses a Legitimate Application of the EU Database
Directive. SCRIPTed 1: 83-118
CRIDES Working Paper Series 2018
Guibault L (2002) Copyright limitations and contracts. Kluwer Law International, The Hague
Guibault L, et al. (2012) Study on the implementation and effect in Member States' laws of
Directive 2001/29/EC on the harmonisation of certain aspects of copyright and related
rights in the information society. Report to the European Commission, DG Internal
Market, February 2007. Amsterdam Law School Research Paper No. 2012-28. Institute
for Information Law Research Paper No. 2012-23. Accessed 18 August
Hall W, Pesenti J (2017) Growing the Artificial Intelligence Industry in the UK.
Accessed 18 August 2018
Harkous H, et al. (2018) Polisis: Automated Analysis and Presentation of Privacy Policies Using
Deep Learning. arXiv preprint arXiv:1802.02561. Accessed 18 August 2018
Hearst MA (2003) Text Data Mining. In: Mitkov R (ed.), The Oxford Handbook of
Computational Linguistics. Oxford University Press, Oxford, New York, 616-62
Helberger N (2013) Forms matter: Informing consumers effectively (Study commissioned by
BEUC). Accessed 18
August 2018
Helberger N, et al. (2017) The perfect match?a closer look at the relationship between eu
consumer law and data protection law. Common Market Law Review 54(5): 1427-1465
Helberger N, Hugenholtz PB (2007) No place like home for making a copy: Private copying in
European copyright law and consumer law. Berkeley Tech. LJ, 22(3): 1061-1093
Helberger N, et al. (2013) Digital content contracts for consumers. Journal of consumer policy
36(1): 37-57
Helleringer G, Sibony A-L (2017) European Consumer Protection through the Behavioral Lens.
Columbia Journal of European Law 23(3): 607-649
Hillman RA (2006) Online Boilerplate: Would Mandatory Website Disclosure of E-Standard
Terms Backfire? Michigan Law Review 104(5): 837-856
Hilty R et al. (2008) Declaration on a balanced interpretation of the “three-step test” in copyright
law. International Review of Intellectual Property and Competition Law 39(6): 707-712
Hilty R, Richter H (2017) Position Statement of the Max Planck Institute for Innovation and
Competition on the Proposed Modernisation of European Copyright Rules Part B
Limitations to text and data mining and consumer empowerment
Exceptions and Limitations (Art. 3–Text and Data Mining).
tement_Part_B_Chapter_1_Update23022017.pdf. Accessed 18 August 2018
Hugenholtz PB (2000) Copyright, contract and code: what will remain of the public domain.
Brook. J. Int'l L. 26: 77-90
Jasiewicz MI (2012) Copyright Protection in an Opt-Out World: Implied License Doctrine and
News Aggregators. The Yale Law Journal 122(3): 837-850
Kästle-Lamparter D (2018) Pre-contractual information duties. In: Jansen N, Zimmermann R
(eds.), Commentaries on European Contract Laws. Oxford University Press, Oxford,
Kretschmer M, et al. (2010) The relationship between copyright and contract law. Accessed 18
August 2018
Lippi M, et al. (2018) CLAUDETTE: an Automated Detector of Potentially Unfair Clauses in
Online Terms of Service. arXiv preprint arXiv:1805.01217. Accessed 18 October 2018
Loos M (2015) Transparency of standard terms under the Unfair Contract Terms Directive and
the proposal for a common European sales law. European Review of Private Law 23(2):
Margoni T (2012) Eccezioni e limitazioni al diritto d'autore in Internet= Exceptions and
Limitations to Copyright Law in the Internet.
Accessed 18 August 2018
Margoni T, Dore G (2016) Why We Need a Text and Data Mining Exception (But it is Not
Enough). Accessed 18 August
Margoni T, Kretschmer M (2018) The Text and Data Mining exception in the Proposal for a
Directive on Copyright in the Digital Single Market: Why it is not what EU copyright
law needs.
directive-digital-single-market-not-what-eu-copyright-needs/. Accessed 18 August 2018
Mazziotti G (2008) EU Digital Copyright Law and the End-User. Springer, Verlag, Berlin,
Micklitz H-W, et al. (2009) Understanding EU consumer law. Intersentia, Antwerp
CRIDES Working Paper Series 2018
Montagnani ML (2007) Dal peer-to-peer ai sistemi di Digital Rights Management: primi appunti
sul melting pot della distribuzione online. Il diritto d’autore (1):1-57
Montagnani ML, Aime G (2017) Il text and data mining e il diritto d'autore. AIDA: 376-394
Myska M, Harasta J (2016) Less is More: Protecting Databases in the EU after Ryanair. Masaryk
UJL & Tech. 10: 170-198
OECD (2016) Protecting Consumers In Peer Platform Markets: Exploring The Issues.
Accessed 18 August 2018
Poort J (2018) Borderlines of Copyright Protection: An Economic Analysis. In: Hugenholtz PB
(ed.), Copyright Reconstructed: Rethinking Copyright’s Economic Rights in a Time of
Highly Dynamic Technological and Economic Change. Wolters Kluwer, Alphen aan den
Rijn, 283-338
Porat A, Strahilevitz LJ (2013) Personalizing Default Rules and Disclosure with Big Data. Mich.
L. Rev. 112(8): 1417-1478
Radin MJ (2013). Boilerplate: the Fine Print, Vanishing Rights, and the Rule of Law. Princeton
University Press, Princeton, Oxford
Rosati E (2018) National and EU text and data mining exceptions: room for coexistence?
Accessed 18 August 2018
Rosati E (2013a). Originality in EU Copyright. Edward Elgar, Cheltenham, UK, Northampton,
Rosati E (2013b) Towards an EU-wide Copyright?(Judicial) pride and (legislative) prejudice.
Intellectual Property Quarterly 1:47-68
Rotenberg B, Compañó R (2009) Search Engines for audio-visual content: Copyright law and its
policy relevance Telecommunication Markets. In: Curwen P, et al. (eds.)
Telecommunication Markets. Springer, Heidelberg and New York, 113-139
Sadeh N, et al. (2013) The usable privacy policy project. http://reports- Accessed 18 august 2018
Scholz T (2016) Platform Cooperativism. Challenging the Corporate Sharing Economy. Rosa
Luxemburg Stiftung, New York
Sire G (2015) Inclusion exclue : le code est un contrat léonin. Enquête sur la valeur technique
et juridique du protocole robots.txt. Réseaux 189(1): 187-214
Limitations to text and data mining and consumer empowerment
Stamatoudi IA (2016) Text and data mining. In: Stamatoudi IA (ed.), New Developments in EU
and International Copyright Law. Wolters Kluwer, Alphen aan den Rijn, 251-282Strowel
A (1993) Droit d'auteur et copyright, Divergences et convergences. Bruylant, Paris,
Strowel A (2007) Google et les nouveaux services en ligne: quels effets sur l’économie des
contenus, quels défis pour la propriété intellectuelle. Journal des tribunaux 22: 589-598
Strowel A (2011). Quand Google défie le droit. Larcier, Bruxelles
Strowel A (2012) European Copyright: Beyond the Additions Made by the European Court of
Justice, Some Pieces Are Still Missing. In: Janssens M-C, Overwalle GV (eds.),
Harmonisation of European IP Law, In Honor of Fr. Gotzen. De Boeck, Bruxelles, 73-
Strowel A (2014) Droit d’auteur et copyright. Convergences des droits, régulation différente des
contrats Mélanges en l'honneur du Professeur André Lucas. LexisNexis, Paris, 699-717
Strowel A (2015) Fair Compensation for Private Copying Copyright and the Digital Agenda for
Europe: Current Regulations and Challenges for the Future. Sakkoulas Publications,
Athens, 189-195
Strowel A (2018) Reconstructing the Reproduction and Communication to the Public Rights:
How to Align Copyright with Its Fundamentals. In: Hugenholtz PB (ed.), Copyright
Reconstructed Rethinking Copyright’s Economic Rights in a Time of Highly Dynamic
Technological and Economic Change. Wolters Kluwer, Alphen aan den Rijn, 203-240
Strowel A, Derclaye E (2001) Droit d’auteur et numérique: logiciels, bases de données,
multimédia. Bruylant, Bruxelles
Sunstein C (2012) Informing Consumers through Smart Disclosure.
through-smart-disclosure. Accessed 28 August 2018
Triaille JP, et al. (2013) Study on the application of Directive 2001/29/EC on copyright and
related rights in the information society.
detail/-/publication/9ebb5084-ea89-4b3e-bda2-33816f11425b. Accessed 18 October
Triaille JP, et al. (2014) Study on the legal framework of text and data mining (TDM).
4a1d-9895-65290705e2a5/language-en. Accessed 18 October 2018
CRIDES Working Paper Series 2018
Valenti R (2007a) Art. 68-bis. In: Ubertazzi LC (ed.), Diritto d’autore. Estratto da L. C. Ubertazzi
“Commetario breve alle leggi su proprietà intellettuale e concorrenza”. CEDAM, Padua,
Valenti R (2007b) Introduzione agli artt. 65-71-quinquies. In: Ubertazzi LC (ed.), Diritto
d’autore. Estratto da L.C. Ubertazzi “Commentario breve alle leggi su proprietà
intellettuale e concorrenza”. CEDAM, Padua, 190-196
Vallés RC (2009) Research Handbook on the Future of EU Copyright The Requirement of
Originality. Edward Elgar, Cheltenham, UK, Northampton, MA, USA
Vila T, et al. (2003) Why we can't be bothered to read privacy policies models of privacy
economics as a lemons market. ICEC '03 Proceedings of the 5th international conference
on Electronic commerce, 403-407
Waelde C, et al. (2013) Contemporary intellectual property: law and policy. Oxford University
Press, Oxford
Walter M, Von Lewinski S (2010) European Copyright Law. A Commentary. Oxford University
Press, Oxford
Wilhelmsson T (2004) The Abuse of the “Confident Consumer” as a Justification for EC
Consumer Law. Journal of Consumer Policy 27(3): 317-337
Limitations to text and data mining and consumer empowerment
Table 1. Overview of limitations and prohibitions in Sharing Economy Platforms (IoP project)1
IP-related CLAUSE
What users can do
What users cannot do
Does the platform use an exclusion protocol
to keep robots away from T&C and privacy
policy webpages?
Access and use the content only for
personal and non-commercial use
- reproduce, modify, prepare
derivative works, etc.
- scraping, data mining
Access and use the content only for
personal and non-commercial use
- reproduce, modify, prepare
derivative works, etc.
- data mine a substantial part of the
- print or download fragments of the
materials appearing on the Website,
for non-commercial purposes,
informative and personal
- copy the material appearing on the
website with the intention to
disseminate and communicate them
within the family circle
- copy, send, distribute, sell, publish,
transmit, circulate, arrange or modify
any Website material (unless
otherwise provided by law)
No provisions TDM-related
The following robots are excluded from the
entire server: BLEXBot, MauiBot,
DomainCrawler, AhrefsBot,
1 This analysis has been primarily conducted within the framework of the project “The Internet of Platforms: an empirical study on private ordering and consumer protection in the sharing
economy” ( All links have been accessed on 10/08/2018.
2 Terms available in French and Dutch only. There are three different contracts: one for Brussels, one for Wallonie, and one for Flanders. The version analysed here is the contract available
for users in Brussels.
CRIDES Working Paper Series 2018
OmniExplorer_Bot/1.09, SemrushBot,
SemrushBot-SA, Mechanics,,
No provisions TDM-related
-use the platform contents for non-
commercial purposes
- not exploit in any way the content
No provisions TDM-related
Access and view the content only for
personal and non-commercial use
- copy, adapt, modify, prepare
derivative works, etc.
- use any robots, spider, crawler,
scraper or other automated means or
processes to access, collect data or
other content from or otherwise
interact with the Airbnb Platform for
any purpose
YES. The “Terms” page is specifically
3 Terms available in French and Dutch only.
Limitations to text and data mining and consumer empowerment
Access and use the contents and the
services only for their intended
- reproduce, modify, prepare
derivative works, etc. (unless
otherwise provided by law)
- data mining robots or similar data
gathering or extraction methods
- download (other than the page
caching) of any portion of the Service
The following robots is excluded from the
entire server: ADmantX
- download, display or print
individual pages of the Site to
evidence any agreement with
HomeAway and to retain a copy of
their bookings.
- The relevant file or the relevant
printout must clearly bear the text
© 2017 HomeAway - All Rights
- reproduce, copy in whole or in part
- use robots, spiders or similar data
gathering and extraction tools
YES. It disallows a directory that contains also
Terms and Privacy Policy
fice space
- store temporary copies of the
website materials in RAM (incidental
to the surfing)
- store files automatically cached by
the Web browser for display
enhancement purposes
- print or download one copy of a
reasonable number of pages of the
Site for your use in connection with
the Services.
- download a single copy of the
application to the personal computer
or mobile device solely for personal,
non-commercial use
- reproduce, distribute, modify,
create derivative works of, publicly
display, publicly perform, republish,
download, store or transmit any of
the material on the Website, with the
exception of the permitted uses
- use manual or automated software,
devices, scripts robots, other means
or processes to access, “scrape,”
“crawl” or “spider” any web pages or
other services contained in the Site,
Application, Services or Content
CRIDES Working Paper Series 2018
No provisions TDM-related
Bar d’Office4
fice space
- obtain or attempt to obtain any
materials or information through any
means not intentionally made
available by the platform
- download and print material from
the Site solely for the purposes of
keeping a record of the Booking and
the Property booked.
- not copy, transmit, modify,
republish, save, pass off, or link to
any content or material on the Site
without the platform prior written
YES. The “Termsandconditions” and
“privacy” pages are specifically disallowed
-reproduce, decompile, reverse
engineers, etc.
Meal Sharing
- Access, view, and print Meal
Sharing and User Content for non-
commercial, at-home use
- reproduce, duplicate, copy, reverse
engineer, sell, resell or exploit any
portion of the Service, use of the
Service, or access to the Service
without the express written
permission by the platform
4 Terms and conditions are named “Disclaimer” and it’s a miscellaneous:
5 The IP provisions are listed in their Privacy policy [sic!]
Limitations to text and data mining and consumer empowerment
- copy, multiply, spread or provide
the design and contents, including
trademarks, logos, images, photos
and texts on the website to others in
any way without written permission
No terms available
No terms available
Menu next
All rights reserved
- use the platform for personal and
non-commercial purposes
- Not copy, modify, alter, adapt, make
available, translate, port, reverse
engineer, decompile, or disassemble
any portion of the Platform in any
- manually or automatically with the
use of any robot, spider, crawler, any
search or retrieval application, or use
other automatic device, process or
method to access the Platform and
retrieve, index and/or data-mine any
6 The platform closed during the performance of the study
7 The platform closed during the performance of the study.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Terms of service of on-line platforms too often contain clauses that are potentially unfair to the consumer. We present an experimental study where machine learning is employed to automatically detect such potentially unfair clauses. Results show that the proposed system could provide a valuable tool for lawyers and consumers alike.
Full-text available
In this Article, we identify whether and how European Union consumer protection rules could incorporate more behavioral wisdom. The relevance of behavioral insights to consumer rights is obvious as consumer protection rules tend to ignore how boundedly rational consumers can be when they make their decisions. The rules are all too often written with a fictional consumer in mind: one who reads labels and checks the terms and conditions. Though the relevance of behavioral insights to consumer protection is universal, the European context exhibits specific features. In Europe, paternalism is rarely seen as a matter of principle. The debate is therefore not whether a behavioral approach can offer minimalist regulatory approaches preserving freedom of choice, or whether it provides evidence that is robust and general enough to justify paternalistic interventions; rather, it is whether and how a more behavioral approach can make EU law more effective and European consumers better off. The focus in this Article is precisely on how behavioral insights are being incorporated and could be incorporated. The question is important since EU consumer law has evolved into an apparent anti-model of behavioral regulation, featuring a much-criticized load of mandatory information requirements. The internal market constraints that still exist should however not be analyzed as preventing a behavioral turn. The question is also timely since new rules are in preparation at the EU level in the field of consumer protection. Can the EU legislator take useful inspiration from the insights developed in behavioral sciences? Various dimensions are studied. First, disclosure mandates are a central feature of EU consumer law that has been severely criticized in light of behavioral findings. Though disclosure mandates are over-used, they can still serve a useful purpose provided their use is adequately streamlined and enhanced with “smart” design. Secondly, advice mandates also suffer from limits inherent to the disclosure of information, in particular the need for consumers to process complex information, but they too can be made more effective. Thirdly, simplification is a behaviorally sound intuition, which has been badly implemented, notably because the EU legislator has erred on what to simplify. Finally, the question whether further developments in behavioral research should primarily be implemented at the level of the Member States or of the EU raises challenges. In any event, the heavy machinery of choice protection should only be deployed for choices that matter to consumers.
Technical Report
Full-text available
The Proposal for a Directive on Copyright in the Digital Single Market (the Proposal) contains a number of provisions intended to modernise EU copyright law and to make it “fit for the digital age”. 1 Some of these provisions have been object of a lively scholarly debate in the light of their controversial nature (the proposed adjustment of intermediary liability for copyright purposes contained in Art. 13, see here at p. 7) or because they propose to introduce a new right within the already variegate EU neighbouring right landscape (i.e. the protection for press publishers contained in Art. 11). Far less attention has attracted the provision contained in Art. 3 of the Proposal dedicated to “Text and data mining” (however, see here and here). The goal of Art. 3 is to introduce a mandatory exception in EU copyright law which will exempt acts of reproduction made by research organisations in order to carry out text and data mining for the purposes of scientific research. In this blog Thomas Margoni and Martin Kretschmer discuss Art. 3 and explain why its formulation – although underpinned by the right innovation policy goal – is wrong.
Full-text available
This article is concerned with the risks associated with the monopolisation of information that is available from a single source only. Although there is a longstanding consensus that sole-source databases should not receive protection under the EU Database Directive, and there are legislative provisions to ensure that lawful users have access to a database’s contents, Ryanair v. PR Aviation challenges this assumption by affirming that the use of non-protected databases can be restricted by contract. Owners of non-protected databases can contractually exclude lawful users from taking the benefit of statutorily permitted uses, because such databases are not covered from the legislation that declares this kind of contract null and void. We argue that this judgment is not consistent with the legislative history and can have a profound impact on the functioning of the digital single market, where new information services, such as meta-search engines or price-comparison websites, base their operation on the systematic extraction and reutilisation of materials available from online sources. This is an issue that the Commission should address in a forthcoming evaluation of the Database Directive. Keywords: Copyright; Database right; Databases; EU law; Price comparisons; Terms and conditions; Websites; Sole-Source databases
This paper discusses whether and to what extent the transparency principle is applicable to standard contract terms legislation under European Union law and what the consequences are when the principle, in so far as it is recognized, is breached. To that extent, it focuses first on the Unfair Contract Terms Directive and second on the proposal for a Common European Sales Law. Résumé: Cet article traite si et dans quelle mesure le principe de transparence s'applique à la législation européenne de conditions générales et quelles sont les conséquences lorsque le principe, dans la mesure où il est reconnu, a été violé. Dans cette mesure, il se concentre d'abord sur la Directive concernant les clauses abusive dans les contrats conclus avec les consommateurs et d'autre part sur la proposition d'un Droit Commun Européen de la Vente. Zusammenfassung: Dieser Aufsatz beschreibt, ob und inwieweit der Grundsatz des Transparenzgebots gilt nach das europäischen AGB-Recht und was die Konsequenzen sind, wenn das Gebot verletzt wird. Der Aufsatz konzentriert sich zunächst auf die Klauselrichtlinie und zweitens über den Vorschlag für einen Gemeinsamen Europäischen Kaufrechts.
Book The protection of the investment made in collecting, verifying or presenting database contents is still not harmonised internationally. Some laws over-protect database contents, whilst others under-protect them. This book examines and compares several methods available for the protection of investment in database creation – namely, intellectual property, unfair competition, contract and technological protection measures – in order to find an adequate type and level of protection. To this effect, the author uses criteria based on a combination of the economics of information goods, the human rights to intellectual property and to information, and the public interest, proposing a model that can be adopted at international and national levels.