This document discusses the Linked Open Data cloud and tools for working with linguistic data as linked open data. It describes several key datasets and ontologies in the LOD cloud relevant to linguistics, including DBpedia, BabelNet, LexVo, and Domain datasets. It also discusses challenges around interoperability between datasets due to differences in granularity, language codes, and URIs. Standards and efforts discussed for addressing these challenges include LexInfo, Linked Open Vocabularies, ISOcat, Glottolog, and OLiA. The document concludes by discussing how to submit new datasets to the LOD cloud and using linked data in natural language processing services and pipelines.

1. This project has received funding from the European Union’s Horizon 2020 research and innovation programme
under grant agreement No 731015.
John P. McCrae1, Thierry Declerck2
Linked Open Data
Cloud
1Insight Centre for Data Analytics, National University of Ireland Galway
2Austrian Centre for Digital Humanities

2. This project has received funding from the European Union’s Horizon 2020 research and innovation

3. Hitchhiker’s guide to the Linked
Open Data Cloud

4. DBpedia
• Largest node in the linked open data cloud
• “Nucleus for a web of open data”
• Most data is derived by parsing Wikipedia
• E.g., https://en.wikipedia.org/wiki/C++
• => http://dbpedia.org/resource/C++
• Uses transparent content negotiation

5. This project has received funding from the European Union’s Horizon 2020 research and innovation
Transparent content negotiationI want to know about “C++” and I understand RDF and
HTML
curl -H "Accept: application/rdf+xml;text/html" -I
http://dbpedia.org/resource/C++
HTTP/1.1 303 See Other
...
Location: http://dbpedia.org/data/C++.xml
...
Go to this location for the RDF/XML version

6. This project has received funding from the European Union’s Horizon 2020 research and innovation
Transparent content negotiationI want to know about “C++” and I only know HTML
curl -H "Accept: text/html" -I
http://dbpedia.org/resource/C++
HTTP/1.1 303 See Other
...
Location: http://dbpedia.org/page/C++
...
Go to this location for the HTML version
Use
/resource/
URL to refer
to concept

7. This project has received funding from the European Union’s Horizon 2020 research and innovation
DBpedia Pages
Links to other
resources
Ontology
properties

8. This project has received funding from the European Union’s Horizon 2020 research and innovation
DBpedia ontologyAxioms
Labels
External
Links

9. This project has received funding from the European Union’s Horizon 2020 research and innovation
WikiData RDF Version:
https://www.wikidata.org/entity/Q2407

10. BabelNet
Dictionary
compiled from
• Wikipedia
• (Open
Mulitlingual)
WordNet
• Wiktionary
• OmegaWiki
• WikiData

11. LexVo
• Assigns URIs to
words (strings in a
language)
• Contains links to
WordNet,
FrameNet etc
• Definitions of ISO
Language Codes

12. Domain datasets
100’s of domain
specific datasets

13. This project has received funding from the European Union’s Horizon 2020 research and innovation
LexInfo
LexVo
BabelNet
DBpedia

14. Reusing URIs

15. Why reuse URIs
• Data interoperability
• Queries work over multiple datasets
• Semantic definitions allows alignments to be reasoned
• (Often) the creators of the URIs have good idea on how data
should be structured

16. Challenges of interoperability
Susanne Penn
The AT
DT
Fulton NP1s
NNP
County NNL1cb
NNP
Grand JJ
NNP
Jury NN1c
Differences in
Granularity
Linguistic
Differences

17. Language codes
fr
en
de
thProblem: 7,000+
languages and more
dialects, but only
262=676 codes br?
br =
Breton

18. ISO Language Codes
fr-CA
Québécois
ang
Anglo-Saxon
pms
Piedmontese
3-Letter codes with
region cover
minority, historical
languages, right?

19. Variability
How to tag this talk?
• en?
• en-Latn? (As it is not written in Cyrillic)
• en-LV or en-428 (As it is presented in Latvia)
• en-GB or en-826 (As is is composed in British
English)
• en-Latn-GB?Region subtags are used to indicate linguistic variations associated
with or appropriate to a specific country, territory, or region. Typically,
a region subtag is used to indicate variations such as regional
dialects or usage, or region-specific spelling conventions. It can also
be used to indicate that content is expressed in a way that is
appropriate for use throughout a region, for instance, Spanish content
tailored to be useful throughout Latin America.
-- RFC 5646

20. Glottolog
• Identifies languoids (language
varieties)
• Uses URLs
http://glottolog.org/resource/languoid/id/queb1247
• More information can be found
by following the link

21. Linked Open Vocabularies
http://lov.linkeddata.es/

22. ISOcat
• Effort to standardize linguistic vocabulary from ISO Technical
Committee
• Standardized “Data Categories” in a “Registry”
• Discontinued in December 2014

23. Problems with ISOcat
According to Schuurman et al.
• Too easy to get a login
• “Out-of-control”
• Entries were copies of other entries
• “People sometimes copied an entry, just in order to
make sure the original owner would not change the
entry without them knowing it”
• Complexity - Too many obligatory and overly
technical fields
As an alternative the CLARIN concept registry is (still)
being introduced.
I. Schuurman, M. Windhouwer, O. Ohren, D. Zeman, CLARIN Concept Registry: The new semantic registry, in
CLARIN 2015 Selected Papers (2015), pp. 62–70

24. LexInfo
• Ontology for “associat[ing] linguistic information
with respect to any level of linguistic description
and expressivity to elements in an ontology”
• Expands OntoLex-Lemon with a set of general
categories

25. LexInfo - Properties and Values
Properties and
open-world
(non-exhaustive)
list of values

26. LexInfo - Verb Frames
Verb frames
with formal
definitions

27. LexInfo - Arguments
Hierarchies of
arguments to be
used in the
frames

28. OLiA
Ontologies of Linguistic Annotation
• Modular architecture for describing
annotation schemes:
• Reference Model: Common
terminology (similar to LexInfo)
• Annotation Model: Describes a
particular annotation scheme
• Linking Model: Describes the
linking between the reference and
annotation

29. GOLD - General Ontology Linguistic
Description
• Quite popular
• Defines many terms
• ‘Loose’ semantics
• Sometimes has range and domains on properties
• Not clear how this fits together

30. Submitting to the LOD Cloud

31. This project has received funding from the European Union’s Horizon 2020 research and innovation
Go to lod-cloud.net

32. This project has received funding from the European Union’s Horizon 2020 research and innovation
Fill in the form

33. Fields
• Identifier
• unique alphanumeric string
• Title
• Full name in English
• Description
• 2-10 sentence description in English
• Full Download
• A link to the complete dataset, ideally as compressed N-
Triples
• SPARQL Endpoint
• If available
• Other Download
• Other formats for download or partial downloads

34. Fields (2)
• Example
• A single resource that resolves
• Keywords
• Domain
• Defines the colour in the diagram
• Website
• Contact Point
• Links
• Number of triples linking to another dataset in the cloud
• Size
• Number of triples in this dataset
• Namespace, DOI, Image (if desired)

35. Stars for metadata
quality
Availability of
resource

36. Services using linked data

37. Service-oriented architectures
• It is implemented a self-contained operation unit.
• It is a black box for its consumers, which only need
to know the interface, not the implementation.
• It may consist of other underlying services.
Interoperability is a significant challenge here

38. Service chains
Translation
DE => EN
Parser (EN)
Sentiment
Analysis (EN)
Often tricky to do in
practice!

39. Issues with service chains
• Services are often components of pipelines without
clear usage to the end user
• The technology readiness level of services is often
quite low, with little documentation or graphical
user interface,
• Services are hard to install often requiring
compiling from source or specialized libraries not
found in major software repositories.

40. Teanga
• RDF and Linked Data to provide
service interoperability
• Docker to enable easy install
and usage
• Attractive Web Front-End
(Bootstrap, AngularJS, NodeJS)
• Graceful control of errors

41. This project has received funding from the European Union’s Horizon 2020 research and innovation

42. This project has received funding from the European Union’s Horizon 2020 research and innovation

43. This project has received funding from the European Union’s Horizon 2020 research and innovation

44. LAPPS Grid
Defines key vocabularies for service interoperability
• LAPPS Interchange Format (JSON-LD)
• Web Service Exchange Vocabulary
• Human-in-the-loop workflow construction using
Galaxy

45. This project has received funding from the European Union’s Horizon 2020 research and innovation

46. Summary

47. Summary
• Linked Open Data Cloud
• Big
• Many relevant tools
• Fragmented
• Interoperability is less terrible than other systems

48. Thanks.
This publication has emanated from research supported in part by a research grant from Science
Foundation Ireland (SFI) under Grant Number SFI/12/RC/2289, co-funded by the European Regional
Development Fund