This document provides methodological guidelines for publishing linked data. It outlines the main steps in the process, including identifying and analyzing data sources, designing URIs, defining licenses, modeling ontologies, transforming and cleansing data, linking datasets, and publishing and discovering the linked data. The guidelines are based on experiences applying these methods to real government datasets.

1. Methodological Guidelines for
Publishing Linked Data
g
Boris Villazón-Terrazas, Oscar Corcho
Facultad de Informática, Universidad Politécnica de Madrid
,
Campus de Montegancedo sn, 28660 Boadilla del Monte, Madrid
http://www.oeg-upm.net
{bvillazon,ocorcho}@fi.upm.es
Phone: 34 91 3366605 Fax: 34 91 3524819
34.91.3366605, 34.91.3524819
Slides available at: http://www.slideshare.net/boricles/
Acknowledgements: Asunción Gómez-Pérez, Luis M. Vilches,
Victor Saquicela, Al
Vi t S i l Alexander d L ó and many others th t we
d de León, d th that
may have omitted.
WorkdistributedunderthelicenseCreativeCommonsAttribution-
Noncommercial-Share Alike 3.0

2. Main References
Wood, David (Ed) Linking Government Data - 2011
Methodological Guidelines for Publishing Government Linked Data
Boris Villazón-Terrazas, Luis M. Vilches, Oscar Corcho, Asunción Gómez-Pérez
Best Practices for Publishing Linked Data
W3C Editor’s Draft – Government Linked Data Working Group
Michael Hausenblas, Bernadette Hyland, Boris Villazón-Terrazas
https://dvcs.w3.org/hg/gld/raw-file/bcb72f87b5cc/bp/index.html
Cookbook for Open Government Linked Data
W3C Editor’s Draft – Government Linked Data Working Group
Bernadette Hyland, Boris Villazón-Terrazas, Sarven Capadisli
http://www.w3.org/2011/gld/wiki/Linked_Data_Cookbook
http://www w3 org/2011/gld/wiki/Linked Data Cookbook

3. Guidelines for Publishing Linked Data
• The process of publishing Linked Data has an
iterative incremental life cycle model.
• Based on our experience in the production of Linked
Data in several Governmental Contexts, have been
applied in real case scenarios.
3

4. 4

5. 5

6. Specification
• Identification and analysis of the data
sources
• URI design
• Definition of the license
6

7. Specification
Identification and analysis of the data sources
We have to distinguish
• O
Open and publish d t th t government agencies h
d bli h data that t i have
not yet opened up and published
• Task that may require contacting to specific government data
owners to get access to their legacy data
• Reuse and leverage on data already opened up and
p
published by g
y government agencies
g
• Task to look for these data in public government catalogs
• Open Government Data
• datacatalogs org
datacatalogs.org
• Open Government Catalog
7

8. Specification
Identification and analysis of the data sources
After we have identified and selected the government data
sources
• Search and compile all the available data and
documentation about those resources
• Identify the schema of those resources including
conceptual components and th i relationships
t l t d their l ti hi
• Identify the items in the domain i e things whose
domain, i.e.,
properties and relations are described in the data
sources
8

9. Specification
GeoLinkedData – Identification of the data sources
Agreement with the IGN
IGN
National Geographic Institute of Spain
Oracle & MySQL
Data
D t sources available
il bl
in a public data catalog
INE
National Statistic Institute of Spain
9

10. Specification
GeoLinkedData – Analysis of the data sources
Year
Province Industry Production Index
10

11. Specification
URI Design
• Use meaningful URIs, instead of opaque URIs, when
possible
• Separate TBox (ontology model) from ABox
(instances) URIs
URIs.
• Base URI
http://data.gov.bo/
http://health.data.gov.bo/
• TBox URIs
http://data.gov.bo/ontology/{class|property}
p g gy { |p p y}
• ABox URIs
http://data.gov.bo/resource/
http://data.gov.bo/resource/province/Tiraque
http://data gov bo/resource/province/Tiraque
11

12. Specification
GeoLinkedData - URI design
• Base URI
http://linkeddata.es/
http://geo.linkeddata.es/
• TBox URIs
http://geo.linkeddata.es/ontology/{concept|property}
http://geo.linkeddata.es/ontology/Provincia
http://geo linkeddata es/ontology/Provincia
• ABox URIs
http://geo.linkeddata.es/resource/{r. type}/{r. name}
http://geo.linkeddata.es/resource/Provincia/Madrid
12

13. Specification
Definition of the license
• Several possibilities
• The UK Open Government License
• Open Database License
• Public Domain Dedication and License
• Open Data Commons Attribution License
• The C
Creative C
Commons Licenses
It is also possible to reuse and apply an existing license
p pp y g
of the government data sources.
13

14. Specification
GeoLinkedData - Definition of the license
• Reusing the original license of the government data
sources. IGN and INE data sources have their own
license, similar t Att ib ti Sh
li i il to Attribution-Share Alik 2 5 G
Alike 2.5 Generic
i
License
http://creativecommons.org/licenses/by-sa/2.5/
14

15. 15

16. Modelling
Ontology
• An ontology is an engineering artifact, which provides:
• A set of terms
• A set of explicit assumptions regarding the intended meaning of the terms.
• Almost always including concepts and their classification
• Almost always including properties between concepts
• Shared understanding of a domain of interest
• Ontologies expressed in OWL or RDF(S), both based on RDF
16

17. Modelling
Reuse available vocabularies
Search f suitable
S h for it bl
vocabularies
Linked Open Vocabularies
are there Yes Build the vocabulary by
suitable reusing available
g
vocabularies? vocabularies
No
…
17

18. Modelling
Reuse available non-ontological resources
Highly reliable Web Sites
Search f suitable
S h for it bl Domain related
Domain-related sites
non-ontological resources
Government Catalogs
are there Yes Build the vocabulary by
suitable transforming available
t f i il bl
resources? resources
No
Build the vocabulary from
scratch
18

19. Modelling
GeoLinkedData
WGS84 Geo
Positioning: an RDF
vocabulary scv:Dimension
scv:Item
scv:Dataset
hydrographical
phenomena (rivers,
lakes, etc.)
Vocabulary for
instants, intervals,
durations, etc.
Names and
international code
Ontology for OGC systems for
Geography Markup territories and
Language groups
Classes 33 33
Object Properties 44 44
Data Properties 318 318
http://neon-toolkit.org/
19

20. Modelling
GeoLinkedData
20

21. 21

22. Generation
• Transformation
• Data cleansing
• Linking
22

23. Generation
Transformation
• Take the data sources selected in the specification
activity and transform them to RDF according to the
vocabulary created i th modelling activity
b l t d in the d lli ti it
• Some tools
• CSV and spreadsheets
• RDF extension of Google Refine, XLWrap, RDF123, NOR2O
• RDB
• D2R Server, ODEMapster, W3C RDB2RDF WG – R2RML
• XML
• GRDDL, ReDeFer
23

24. Generation
GeoLinkedData - Transformation
NOR2O
INE
ODEMapster
IGN
Geospatial Geometry2RDF
column
IGN
24

25. Generation
GeoLinkedData - Transformation
Industry Production Index Year
Province
NOR2O
25

26. Generation
GeoLinkedData - Transformation
• R2O is an e te s b e, fully dec a at e language to desc be
s a extensible, u y declarative a guage describe
mappings between relational database schemas and ontologies.
• The ODEMapster processor generates RDF instances from
relational instances based on the mapping description
pp g p
expressed in the R2O document
www.oeg-upm.net/index.php/en/downloads/9-r2o-odempaster
26

27. Generation
GeoLinkedData - Transformation
• Creation of the R2O Mappings
27

28. Generation
GeoLinkedData - Transformation
Excerpt of the R2O document
28

29. Generation
GeoLinkedData - Transformation
• Tool for generating RDF from geometrical information
• The geometry could be available in GML or WKT
• The RDF generated follows our Geometry Model
http://www.oeg-upm.net/index.php/en/downloads/151-geometry2rdf
29

30. Generation
GeoLinkedData - Transformation
Oracle STO UTIL package
SELECT TO_CHAR(SDO_UTIL.TO_GML311GEOMETRY(geometry))
AS Gml311Geometry
FROM "BCN200"."BCN200_0301L_RIO" c
WHERE c.Etiqueta='Arroyo'
30

31. Generation
GeoLinkedData - Transformation

32. Generation
Data Cleansing
• To find possible errors, identified by Hogan et al.
• http-level issues, such as accessibility and derefencability,
e.g.,
e g HTTP URIs ret rn 40 /50 errors
return 40x/50x
• reasoning issues such as namespace without vocabulary,
e.g., rss:item term invented
• malformed/incompatible datatypes, e.g., “true” as xsd:int
• To fix the identified errors
32

33. Generation
GeoLinkedData – Data Cleansing
• Errors
• Some resources, with the same name, were mixed. For
example,
e ample Granada municipality belongs to Granada
m nicipalit
province, and La Granada municipality belongs to Barcelona
Province.
• Autonomous communities that only have one province, e.g.,
Murcia Region, missed some municipalities, but their
corresponding provinces, e g Murcia Province have the
provinces e.g., Province,
correct number of municipalities.
• S
Some hydrographical resources missed some parts of their
f
geometrical information.
33

34. Generation
Linking
Identify suitable data sets http://ckan.net
as linking targets
Discover relationships
between data items
LIMES Silk Framework
http://aksw.org/Projects/limes http://www4.wiwiss.fu-berlin.de/bizer/silk/
Validate the relationships
discovered sameAs Validator
http://oegdev.dia.fi.upm.es:8080/sameAs/
34

35. Generation
GeoLinkedData - Linking
GeoLinked
Data
DBPedia GeoNames
…. …. ….
http://dbpedia.org/re http://geo.linkeddata http://sws.geoname
source/Madrid .es/.../Madrid s.org/6355233/
…. …. ….
35

36. Generation
GeoLinkedData - Linking
http://oegdev.dia.fi.upm.es:8080/sameAs/
http://oegdev dia fi upm es:8080/sameAs/
36

37. 37

38. Publication
• Dataset publication
• Metadata publication
• Dataset discovery
38

39. Publication
Dataset Publication
• Tools for storing RDF
• Virtuoso Universal Server, Jena, Sesame, 4Store, YARS,
OWLIM
• SPARQL endpoint and Linked Data frontend
• Pubby, Talis Platform, Fuseki
39

40. Publication
Metadata Publication
• VoID allows to express metadata about RDF
datasets
• Open Provenance Model
40

41. Publication
Dataset discovery
• Register the dataset into CKAN Registry
• Generate sitemap files for your dataset, by using
sitemap4rdf
• Submit the sitemap location to Google and Sindice
http://www.w3.org/wiki/TaskForces/CommunityProjects/LinkingOpenData/DataSets/CKANmetainformation
41

42. Publication
GeoLinkedData – Dataset publication
HTML Linked Data SPARQL
Including Provenance Pubby
Support
http://www4.wiwiss.fu-berlin.de/pubby/ Pubby 0.3
Virtuoso 6.1.0
610
42

43. Publication
GeoLinkedData – Dataset discovery
43

44. 44

45. Exploitation
Streaming resources
45

46. Exploitation
GeoLinkedData
http://oegdev.dia.fi.upm.es/projects/map4rdf/
map4rdf:
• Google maps viewer of RDF resources
• Resources with spatial information
• Extensible with google plugins
• Used in other applications like Aemet Goodrelations
Aemet,
map4rdf SPARQL
Triplestore
46

47. DEMO
http://geo.linkeddata.es/browser
47

48. Provinces
48

49. Capital of Province
49

50. Provinces – Industry Production Index
50

51. Beaches
51

53. Methodological Guidelines for
Publishing Linked Data
g
Boris Villazón-Terrazas, Oscar Corcho
Facultad de Informática, Universidad Politécnica de Madrid
,
Campus de Montegancedo sn, 28660 Boadilla del Monte, Madrid
http://www.oeg-upm.net
{bvillazon,ocorcho}@fi.upm.es
Phone: 34 91 3366605 Fax: 34 91 3524819
34.91.3366605, 34.91.3524819
Slides available at: http://www.slideshare.net/boricles/
Acknowledgements: Asunción Gómez-Pérez, Luis M. Vilches,
Victor Saquicela, Al
Vi t S i l Alexander d L ó and many others th t we
d de León, d th that
may have omitted.
WorkdistributedunderthelicenseCreativeCommonsAttribution-
Noncommercial-Share Alike 3.0