Licensing Linked Data

Licensing Linked DataWorkshop

I-SEMANTICS 2013 Conference

September 6, 2013

Graz / Austria

Tassilo Pellegrini

firstname.lastname[at]fhstp.ac.at

http://de.slideshare.net/pellegrinit/licensing-linked-data

http://de.slideshare.net/pellegrinit/licensing-linked-data

Introductory Statement: Challenges of LinkedData Licensing• Licensing has been widely neglected in Linked Data R&D

• Data licensing is not a trivial issue – especially under conditions of dual licensing• Requires technological knowledge• Requires asset diversification awareness & strategy• Depends on business strategy & models• Is confronted with competing legal regimes (i.e. EU vs. USA)

• Data licensing shapes social relationships by granting and restricting access toresources.

• (Linked) Data licensing defines the access conditions under which transactionswill be performed in the future (by machines).

• Exposing licensing information as Linked Data is the precondition for automatedrights clearance & brokering systems.

Prof. Dr. Tassilo Pellegrini, University of Applied Sciences St. Pölten, Licensing Linked Data

2

Overview

1. The Economic Rationale of Linked Data

2. Creating Licensing Policies for Linked Data

3. Mapping Licenses to Business Models

4. Conclusion


3

The Economic Rationale ofLinked Data

Metadata ShiftResearch Area Pre-Web Post-Web

Metadata Applications / Uses -- 16 %

Cataloging / Classification 14 % 15 %

Classifying Web Information -- 14 %

Interoperability -- 13 %

Machine Assisted Knowledge

Organization

14 % 12 %

Education 7 % 7 %

Digital Preservation/ Libraries -- 7 %

Thesauri Initiatives 7 % 5 %

Indexing / Abstracting 29 % 4 %

Organizing Corporate or Business

Information

-- 4 %

Librarians as Knowledge

Organizers of the Web

-- 2 %

Cognitive Models 29 % 1 %

Research Areas in Library and Information Science (Source: Saumure, Kristie; Shiri, Ali (2008). Knowledge organization trends in library and information studies: a preliminary comparison of pre- and post-web eras. In: Journal of Information Science, 34/5, 2008, p. 651–666)

Prof. Dr. Tassilo Pellegrini, University of Applied Sciences St. Pölten, Licensing Linked Data 5

The survey illustrates four trends:

1) the spectrum of research areas has broadened significantly;

2) certain areas have kept their status over the years (i.e. Cataloging & Classification or Machine Assisted KnowledgeOrganization),

3) new areas of research have entered the discipline (i.e. Metadata Applications & Uses, Classifying Web Information, Interoperability Issues) and others have declined ordissolved into other areas;

4) metadata issues have significantly increased in importancein terms of the quantity of papers that is explicitly andimplicitly dealing with corresponding issues.

Content-Assets

Metadata-Assets

Information Load

Ec

on

om

icR

ele

va

nc

e

Source: Haase, Kenneth (2004). Context for Semantic

Metadata.

In: MM’04, October 10–16, 2004, New York, New York,

USA. ACM

Price Waterhouse Coopers (2009). Technology

Forecast: Spinning a Web of Data. Spring 2009


6

Metadata as a Network Good

„The Value of Metadata rises as the product of the log of the corpus size and the log of the size of the usercommunity increases.“ (Kenneth Haase, 2004)

Metcalfe`s Law

Data in the Content Value Chain

Content

Acquisition

Content

Editing

Content

Bundling

Content

Distribuiton

Content

Consumption

Harvesting, storage & integration ofinternal orexternal datasources forpurposes likeContent Pooling

Semanticanalysis, adaptation & linking of datafor purposeslike Content Enrichment

Contextualisation& perso-nalisationof informationproducts forpurposes likeLanding Pages, Dossiers orCustomizedDelivery

Provision ofmachine-readable& semanticallyinteroperable data& metadata via APIs or Endpoints

Improvedfindability, navigability & visualization on top of semanticmetadata via Semantic Search& Recommenda-tion Engines


7

Pellegrini, Tassilo (2012). Semantic Metadata in the News Production Process. Achievements and Challenges. In: Lugmayr, Artur; Franssila, Heljä; Paavilainen, Janne; Kärkkäinen, Hannu (Eds). Proceeding of the 16th International Academic MindTrek Conference 2012, Tampere / Finland. ACM SIGMM, p. 125-133

Data Traffic Patterns

Source: Andreas Blumauer, Semantic Web Company, 2011Prof. Dr. Tassilo Pellegrini, University of Applied Sciences St.

Pölten, Licensing Linked Data8

Creating Licensing Policiesfor Linked Data

Licenses on the LOD Cloud – State of the Art

License Number of

Datasets

License Not Specified 251

Creative Commons Attribution 135

Creative Commons CCZero 72

Creative Commons Attribution Share-Alike 71

Creative Commons Non-Commercial (Any) 49

Other (Attribution) 38

UK Open Government Licence (OGL) 36

Open Data Commons Open Database License (ODbL) 28

Open Data Commons Public Domain Dedication and Licence (PDDL) 27

Other (Not Open) 26

Other (Open) 25

Other (Public Domain) 25

Open Data Commons Attribution License 14

GNU Free Documentation License 9

Other (Non-Commercial) 9

ukcrown-withrights 6

W3C 1

apache 1

gpl-2.0 1

gpl-3.0 1

Lic

ense

son t

he

LO

D C

loud

(S

ourc

e: P

elle

grin

i&

Erm

ilov

201

3 …

to

appear)

Prof. Dr. Tassilo Pellegrini, University of Applied Sciences St. Pölten, Licensing Linked Data10

1) Licensing has long been neglected, but awareness is rising

2) High heterogeneity of licenses (CC, ODC, GPL, APACHE, individual licenses …)

3) Insufficient / unappropriate protection ofintellectual assets (not all asset types arecovered)

4) The „meaning“ of the various licenses staysimplicit (not machine-readable) – source oferrors & legal uncertainty

A community discussion & standardizationprocess is required to nuture a licensing culturefor Linked Data

See also Prateek et al. (2013): There is no money in LOD (http://knoesis.wright.edu/faculty/pascal/pub/nomoneylod.pdf)

http://knoesis.wright.edu/faculty/pascal/pub/nomoneylod.pdf

Why Linked Data Licensing Matters?

• Data is an intellectual asset and can be protected by intelllectualproperty rights

• Licenses secure (y)our property rights – for private and publicpurposes!

• Licenses create a secure business environment

• Licenses are an efficient means to diversify business models

• Dual Licensing can be used to extend traditional copyright and allowto reuse, share and consume data for purposes not originallyintended


11

Protecting Data as Intellectual Property

Legal Protection Instruments

Copyright Database

Right

Unfair

Practice

Patents

Linked

Data

Assets

Instance Data Case by Case yes yes Case by Case

Metadata Case by Case yes yes Case by Case

Ontology yes yes yes Case by Case

Content yes no yes no

(Services) yes no yes yes

(Technology) yes no yes yes

Prof. Dr. Tassilo Pellegrini, University of Applied Sciences St. Pölten, Licensing Linked Data 12

Pellegrini, Tassilo (2012). Semantic Metadata in the News Production Process. Achievements and Challenges. In: Lugmayr, Artur; Franssila, Heljä; Paavilainen, Janne; Kärkkäinen, Hannu (Eds). Proceeding of the 16th International Academic MindTrek Conference 2012, Tampere / Finland. ACM SIGMM, p. 125-133

Legend:

Copyright … protects theoriginality of creative works.

Database Right … protects the investment made in compiling a database, even when this does not involve the 'creative' aspect that is reflected by copyright.

Unfair Practices Act … protects against fraud, misrepresentation, and oppressive or unconscionable acts or practices by businesses.

Patents … protects a novel solution to a specific technological problem.

Components of a Linked Data Licensing Policy

A Linked Data licensing policy should consist of three components: a machine-readable statement about content-related assets (copyright), a machine-readable statement about database-related assets (database right) and a human-readable Community Norm.

• Herein the contents of a linked dataset, which are comprised of the terms, definitions and its ontological structure, are protected by copyright (or Creative Commons).

• The underlying database, which is comprised of all independent elements and works that are arranged in a systematic or methodological way and are accessible by electronic or other means, are protected by database right (or Open Data Commons).

• The Community Norm explicitly defines the expectations of the rights holder towards “good conduct” when a dataset is being utilized.


13

Benefits & Limitations of traditional Copyright / Datebase Right• Benefits:

• Easy to handle: rights are usually granted automatically at the moment of publication

• Internationally established institutions & experience of conduct (legal affairs, trials etc.)

• Strong property rights are often the foundation of established business models

• Limitations:• Very restrictive – not suiteable to generate network effects or open innovation

• Regional differences in legal issues (USA vs. Europe)

• Costly & risky to diversify the IPR strategy (i.e. error prone process, learning curves, fears to„let go“)

• Hard to enforce


14

Alternative Protection Instruments I: Creative Commons


15

Creative Commons is an extension to copyright which allows various degrees of freedom to repurpose content via granularly defined constraints. The various licenses can be ordered within a hierarchy of restrictions depending on the usage rights and associated permissions granted by the specific license.

• Benefits:• Enables fine granular expression of usage rights• Allows diversification of creation & distribution of assets• Allows diversification of business models• Contributes to the public domain

• Limitations:• Complex to handle• Might interfere with etsablished business models• Requires cultural change• Hard to enforce

Alternative Protection Instruments II: Open Data Commons


16

Open Data Commons are an extension of Database Right and work analogue to Creative Commons. The various licenses can be ordered within a hierarchy of restrictions depending on the usage rights and associated permissions granted by the specific license.

• Benefits:• Enables fine granular expression of usage rights• Allows diversification of creation & distribution of assets• Allows diversification of business models• Contributes to the public domain

• Limitations:• Very new instrument – work in progress / little experience• Might interfere with etsablished business models• Requires cultural change• Hard to enforce

Community Norm I• Beside licensing information expressed by Copyright / Creative Commons and

Database Right / Open Data Commons a so called Community Norm is the third component of a Linked Data licensing policy.

• A community norm is basically a human-readable recommendation of how the data should be used, managed and structured as intended by the data provider. It should provide administrative information (i.e. creator, publisher, license and rights), structural information about the dataset (i.e. version number, quantity of attributes, types of relations) and recommendations for interlinking (i.e. preferred vocabulary to secure semantic consistency).

• Community norms can differ widely in depth and complexity.


17

Community Norm II: Examples


18

http://www.embeddedmetadata.org/embedded-metatdata-manifesto.php

http://www.embeddedmetadata.org/embedded-metatdata-manifesto.php

Rights Expression Languages I: ODRL

• Rights Expression Languages are used to express usage rights about a digital asset in a machine-readable way.

• A prominent example is ODRL (Open Digital Rights Language), an XML vocabulary to express rights, rules, and conditions - including permissions, prohibitions, obligations, and assertions - for interacting with online content. See: http://www.w3.org/community/odrl/

• ODRL utilizes an Entity-Attribute-Value Model to express a policy about rights and restrictions associated with a digital artefact.

• BUT: ODRL does not provide a licensing attribute. This must be added by referring to other vocabularies like CCREL.

• There are several possibilities how to provide the licensing information: • as an annotation of the HTML document using RDFa, • as a complementary document, which reflects the information on the page for machines (RDF/XML, N3, Turtle

or other notation), • as a public SPARQL endpoint, which can be queried by applications and users, • as a dump file.


19

http://www.w3.org/community/odrl/

Rights Expression Languages II: CCREL

• The Creative Commons Community has developed CCREL (Creative Commons Rights Expression Language) to represent the various CC licenses in a machine-readable format. See http://www.w3.org/Submission/CCREL/ or http://creativecommons.org/schema.rdf

• CCREL complements the ODRL vocabulary. It provides a condensed and hierarchically ordered set of properties that define the actions allowed with certain licenses. These properties can be seamlessly integrated into the ODRL vocabulary and allow to define fine-granular usage policies and constraints associated with a certain asset.

• A combination of ODRL and CCREL is not obligatory. The semantic expressivity of CCREL is sufficient to simply annotate existing assets with licensing information for automated processing. But in case of very complex and differentiated usage scenarios a combination of ODRL and CCREL is recommended, as ODRL provides the necessary semantic expressivity to define fine-granular usage policies associated with a certain asset that go beyond the simple explication of licensing information, i.e. for various user groups or stakeholders.


20

http://www.w3.org/Submission/CCREL/

http://creativecommons.org/schema.rdf

Rights Expression Languages III: CCREL Examples• One RDF triple is enough to attach license information to the work, given

that the license URI is dereferenceable and described by RDF vocabulary provided by Creative Commons Foundation. Here is a basic example of how the CC-BY license can be attached to the asset (ex:myImage):

• @prefix ex: <http://example.org/>.• @prefix cc: <http://creativecommons.org/ns#>.• ex:myImage cc:license <http://creativecommons.org/licenses/by/3.0/> .

• Such an RDF document usually complements an asset (an image in our case) on a web page, where the licensing information should be represented in a human-readable fashion (i.e. with HTML). Via the RDF link an application can attain the information necessary for telling its user how this asset can be processed.


21

1 @prefix xml: <http://www.w3.org/XML/1998/namespace>.

2 @prefix cc: <http://creativecommons.org/ns#>.

3 @prefix foaf: <http://xmlns.com/foaf/0.1/>.

4 @prefix dc: <http://purl.org/dc/elements/1.1/>.

5 @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

6 @prefix dcq: <http://purl.org/dc/terms/>.

7 <http://creativecommons.org/licenses/by/3.0/> cc:legalcode

<http://creativecommons.org/licenses/by/3.0/legalcode>;

8 cc:licenseClass <http://creativecommons.org/license/>;

9 cc:permits cc:DerivativeWorks,

10 cc:Distribution,

11 cc:Reproduction;

12 cc:requires cc:Attribution,

13 cc:Notice;

14 dc:creator <http://creativecommons.org>;

15 dc:identifier "by";

16 dc:title "${Attribution} 3.0 ${Unported}"@i18n,

...

108 dcq:hasVersion "3.0";

109 a cc:License;

110 foaf:logo <http://i.creativecommons.org/l/by/3.0/80x15.png>,

111 <http://i.creativecommons.org/l/by/3.0/88x31.png>.

Rights Expression Languages IV: CCREL Examples• Each RDF license includes the

necessary information encoded in RDF, such as what is allowed and what is prohibited. For example, the CC-BY-SA 3.0 used in the example is represented as follows:

• The code of the CC-BY license defines its URI, legal code, title and other attributes.

• The most important properties of this license are stated on lines 9 - 13: an asset under this license can be distributed, reproduced and made derivation from (cc:permits) if notice, sharealike and attribution are provided (cc:requires).


22

Rights Expression Languages V: ODC Examples• In contrast to Creative Commons, who have

provided CCREL as a machine readable language to express licensing information, ODC licenses are available as plain text only and thus not easily processable by machines.

• But as ODC shares several attributes and characteristics with CC it is possible and reasonable to apply attributes from the CCREL vocabulary.

• On the right you see an example how to combine ODC licensing information with CCREL expressions (lines 7 - 11). Herein the description of the license inside the dataset about a database is the same as in the previous CCREL example.


23

1 @prefix xml: <http://www.w3.org/XML/1998/namespace>.

2 @prefix cc: <http://creativecommons.org/ns#>.

3 @prefix foaf: <http://xmlns.com/foaf/0.1/>.

4 @prefix dc: <http://purl.org/dc/elements/1.1/>.

5 @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

6 @prefix dcq: <http://purl.org/dc/terms/>.

7 @prefix ex: <http://example.org/>

8 ex:myDatabase

9 cc:attributionName "Name of the author"^^xsd:string;

10 cc:attributionURL <http://firstname.lastname.me/>;

11 cc:license <http://opendatacommons.org/licenses/by/1.0/>.

12 <http://creativecommons.org/licenses/by/3.0/> cc:legalcode

<http://creativecommons.org/licenses/by/3.0/legalcode>;

13 cc:licenseClass <http://creativecommons.org/license/>;

14 cc:permits cc:DerivativeWorks,

15 cc:Distribution,

16 cc:Reproduction;

17 cc:requires cc:Attribution,

18 cc:Notice;

19 dc:creator <http://creativecommons.org>;

20 dc:identifier "by";

21 dc:title "${Attribution} 3.0 ${Unported}"@i18n,

...

113 dcq:hasVersion "3.0";

114 a cc:License;

115 foaf:logo <http://i.creativecommons.org/l/by/3.0/80x15.png>,

116 <http://i.creativecommons.org/l/by/3.0/88x31.png>.

MappingLicenses to Business Models

Instance DataMetadata

OntologyContent

ServicesTechnology

Stakeholders

RevenueModel

LinkedData Assets

Linked Data Business Cube

Subsidies

Subscription

Advertising

Certification

Affiliate Program

Value Add

Traffic / SEO

Branding

Revenue Model Legend:Subscription: Selling data & services accessAdvertising: Sell paid placements / advertisementsinside data feeds & servicesCertification: Charge for reviews, verification, compliance checks, quality assuranceAffiliate Program: Charge for affiliate links within datafeeds or servicesValue Add: Utilizing Linked Data to enhance data sets & servicesTraffic / SEO: Utilizing Linked Data to improvefindability & generate trafficBranding: Provide data sets, vocabs & ontologies toshape market & fuel data driven applicationsSubsidies: Public / non-profit funding & regulatorypublishing policies

(Adopted from Brinkner (2010): http://chiefmartec.com/2010/01/the-8th-linked-data-business-model/)

Stakeholder Legend:Internal … within a company // Partners … Between strategic partners // B2B … Business to Business // B2G … Business to Government // B2C … Business toCustomer // C2C … Customer to Customer / Co2Co … Community to Community

http://chiefmartec.com/2010/01/the-8th-linked-data-business-model/

Mapping Licenses to Business Models – A Discussion Proposal


26

Instance Data

Metadata Ontology Content Services Technology / App

Subsidies CC / ODC ODC CC / ODC CC CC / FOSS / ToT FOSS / ToT

Branding CC / © / ODC / DBR ODC CC / ODC CC / © CC / FOSS / ToT FOSS / © / ToT

Traffic / SEO CC / © / ODC / DBR ODC CC / ODC CC / © CC / FOSS / ToT n.r.

Value Add CC / © / ODC / DBR ODC / DBR CC / ODC CC / © CC / © / FOSS / ToT n.r.

Affiliate Prog. CC / © / ODC / DBR ODC / DBR CC / © / ODC / DBR CC / © CC / © / FOSS / ToT FOSS / © / ToT

Certification CC / © / ODC / DBR ODC / DBR CC / © / ODC / DBR CC / © © / ToT © / ToT

Advertising CC / © / ODC / DBR ODC / DBR CC / © / ODC / DBR CC / © © / ToT © / ToT

Subscription © / DBR ODC / DBR CC / © / ODC / DBR © © / ToT © / ToT

Legend:CC … Creative Commons // ODC … Open Data Commons // © … Copyright // DBR … Database Right // FOSS … Free & Open Source License // ToT … Terms of Trade // n.r. … not relevant

Mapping Licenses to Stakeholders – A Discussion Proposal


27

Instance Data

Metadata Ontology Content Services Technology / App

Internal © / DBR DBR © / DBR © © / ToT © / ToT

Partners CC / © / ODC / DBR ODC / DBR CC / © / ODC / DBR CC / © © / ToT FOSS / © / ToT

B2B CC / © / ODC / DBR ODC CC / ODC CC / © CC / FOSS / ToT FOSS / © / ToT

B2G CC / © / ODC / DBR ODC CC / ODC CC / © CC / © / FOSS / ToT FOSS / © / ToT

B2C CC / © / ODC / DBR ODC CC / © / ODC / DBR CC / © CC / © / FOSS / ToT FOSS / © / ToT

C2C CC / © / ODC / DBR ODC / DBR CC / © / ODC / DBR CC / © CC / © / FOSS / ToT FOSS / © / ToT

Co2Co CC / ODC ODC CC / ODC CC CC / FOSS CC / FOSS


Linked Data Licensing – How others do it …*


28

Instance Data Metadata Ontology Content Services Technology / App

BBC (Sports)

© / DBR CC-BY 3.0 CC-BY 3.0 © ?? ??

NYT (Subject Headings)

CC-BY 3.0 CC-BY 3.0 CC-BY 3.0 © ToT n.r.

The Guardian (Music Albums)

?? ?? ?? ?? ToT n.r.

DBpedia CC-BY-SA 3.0 / GNU FDL CC-BY-SA 3.0 / GNU FDLCC-BY-SA 3.0 / GNU

FDLCC-BY-SA 3.0 / GNU

FDLMisc. Misc.

MusicBrainz CC0 CC0 CC0 CC-BY-NC-SA 3.0 CC-BY-NC-SA 3.0 GPLv2

GeoNames CC-BY 3.0 CC-BY 3.0 CC-BY 3.0 CC-BY 3.0 ToT ToT


* Please consider that these licensing policies have regional limitations due to differing regulatory regimes!

Conclusion: Challenges of Linked Data Licensing• Linked Data Licensing is technologically simple, but business-wise complex.

• Linked Data Licensing is a context sensitive issue and requires a goodunderstanding of the intersections of technology, law and business development

• Assets & stakeholders• Markets & ressources• Regulatory & legal conditions• Technology & infrastructure

• Linked Data Licensing challenges traditional business models & culture … can beconsidered a „radical innovation“

• FUTURE: Linked Licensing Data will bring about new applications & services forrights clearance, publishing & billing purposes ... High transformation potential for ecommerce & procurement!


29

Licensing Linked Data

Technology

web of data

linked data rd data

licensing linked data

interoperable data metadata

licensing information

external data sources

licensing policies

semantic metadata