1. Date: 19/10/2012
The Landscape of Ontology Reuse
in Linked Data
María Poveda, Mari Carmen Suárez-Figueroa,
Asunción Gómez-Pérez
Ontology Engineering Group. Departamento de Inteligencia Artificial.
Facultad de Informática, Universidad Politécnica de Madrid.
Campus de Montegancedo s/n.
28660 Boadilla del Monte. Madrid. Spain
{mpoveda, mcsuarez, asun}@fi.upm.es

2. The Landscape of Ontology Reuse in Linked Data 2
Table of contents
• Introduction
• Experimental Method
• Results, Analysis, and Discussion
• Conclusions and Future Works

3. Introduction (i)
3
The Linked Data (LD) initiative enables the easy exposure, sharing, and connecting of data on the Web.
Linked Data principles (http://www.w3.org/DesignIssues/LinkedData.html):
• Use URIs as names for things
• Use HTTP URIs so that people can look up those names.
• When someone looks up a URI, provide useful information, using the standards (RDF, SPARQL)
• Include links to other URIs, so that they can discover more things.
The Landscape of Ontology Reuse in Linked Data

4. Introduction (ii)
4The Landscape of Ontology Reuse in Linked Data
How should I
reuse elements
or vocabularies?
Should I import
another ontology?
Should I reference other
ontology element URIs?
... replicating manually the URI?
... modularizing and
merging ontologies?

5. The Landscape of Ontology Reuse in Linked Data 5
Table of contents
• Introduction
• Experimental Method
• Results, Analysis, and Discussion
• Conclusions and Future Works

6. Experimental Method (i)
6The Landscape of Ontology Reuse in Linked Data
Definitions
Elements
appearing in
a vocabulary.
Local elements: those
defined in the vocabulary
namespace.
External elements:
those not defined in the
vocabulary namespace.
Imported elements: those defined in any of the imported vocabularies
namespaces.
Referenced elements: those not defined in any of the imported
vocabularies namespaces but referenced in the vocabulary being analized.
Referenced by import elements: those not defined in any of the
imported vocabularies namespaces but referenced in at least one of them.
Should I import
another ontology?
Should I reference other
ontology element URIs?
... replicating manually the URI?
... merging ontologies?
Let’s see how others
are reusing terms.

7. 7The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for
rdf, owl, etc
files
Harvesting vocabularies
Experimental Method (ii)
Dataset
(vocabularies to
be analyzed)
Static statistics
Reuse metrics
and reuse
landscape
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance
Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)

8. The Landscape of Ontology Reuse in Linked Data 8
Table of contents
• Introduction
• Experimental Method
• Results, Analysis, and Discussion
• Conclusions and Future Works

9. 9The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for
rdf, owl, etc
files
Harvesting vocabularies
Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)
Results, Analysis, and Discussion (i)
265 vocabulary
prefixes and
namespaces
retrieved from
LOV
242 files
downloaded
52 failed
190 successfully
loaded into
JENA
23 no file
downloaded
56 files
downloaded
manually
6 successfully
loaded into
JENA
Dataset of
196
vocabularies
to be
analyzed
Ontologies difficult to find even manually
looking for them
Not reachable due to connection
problems
ease the task of finding and
understanding the vocabularies for
other developers by providing user
friendly web sites where both the
ontology and its documentation are
easily accessible
ease the tasks of accessing and
processing vocabularies
programmatically by implementing
recommended methods for
publishing vocabularies
http://www.w3.org/TR/swbp-vocab-pub/
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance

10. 10The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for
rdf, owl, etc
files
Harvesting vocabularies
Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance
Results, Analysis, and Discussion (ii)
Classes Object Properties Datatype Properties Total
Locally Defined 5384 3956 1714 11054
Imported 1671 2297 1084 5052
Referenced 783 314 266 1363
ReferencedByImport 488 484 148 1120
Total 8326 7051 3212 18589
59.47% (11054 out of 18589)
original definitions
40.53% (7535 out of 18589)
reused elements
67.05% (5052 out of 7535)
imported elements
18.09% (1363 out of 7535)
referenced elements
14.86% (1120 out of 7535)
referenced by import elements
It could be due to the owl:imports statements mechanism and its transitivity

11. Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)
11The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for
rdf, owl, etc
files
Harvesting vocabularies
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance
Results, Analysis, and Discussion (iii)
Reused ontology Prefix
#being
referenced
http://xmlns.com/foaf/0.1/ foaf 43
http://purl.org/dc/terms/ dc 26
http://www.w3.org/2003/01/geo/wgs84_pos geo 25
http://purl.org/dc/elements/1.1/ dce 14
http://www.w3.org/2004/02/skos/core skos 14
http://www.w3.org/2000/10/swap/pim/contact con 11
http://schema.org/ schema 8
http://purl.org/NET/c4dm/event.owl# event 7
http://dbpedia.org/ontology/ DBpedia* 5
http://purl.org/ontology/bibo/ bibo 5
http://purl.org/vocab/frbr/core# frbr 5
Prefixes marked with an * in this table refer to ontologies that are not included in LOV.
Imported ontology Prefix
#being
imported
http://purl.org/dc/elements/1.1/ dce 15
http://www.w3.org/2003/06/sw-vocab-status/ns vs 10
http://purl.org/dc/terms/ dc 9
http://xmlns.com/foaf/0.1/ foaf 9
http://purl.org/NET/c4dm/event.owl event 8
http://purl.org/goodrelations/v1 gr 5
http://www.w3.org/2006/time time 5
http://purl.org/vocab/vann/ vann 4
http://purl.org/NET/scovo scovo 3
http://purl.org/ontology/ao/core ao 3
http://purl.org/ontology/similarity/ sim 3
http://www.linkedmodel.org/schema/vaem vaem 3
34.69% (68 out of 196) of the
vocabularies use the
owl:imports statement
165 owl:imports statements
53.06% (104 out of 196) of the
vocabularies reference to
other vocabularies
Even though ontology editors support owl:imports
through few simple user interactions while reusing part
of an ontology involves more complex activities (e.g:
module extraction, partitioning, pruning, merging, etc.).

12. 12The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for
rdf, owl, etc
files
Harvesting vocabularies
Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)
Results, Analysis, and Discussion (iv)
0
20
40
60
80
100
120
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1
ReuseRatio
ImportRatio
ReferenceRatio
ReferenceByImportRatio
101 ontologies present a reuse percentage between
0.0 and 0.1
most of the ontologies do little or no reuse
The trend is to adopt a type of reuse for each
ontology, either based on owl:imports statements or
based on referencing element URIs. It is scarce to find
ontologies combining both types of reuse at the same
level.
For those cases with a reuse ratio higher than 60% the
tendency is to achieve this level by importing
ontologies. It could be due to the owl:imports statements
mechanism that include and its transitivity.
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance

13. 13The Landscape of Ontology Reuse in Linked Data
Automatically:
SPARQL, JEN
A API, Vapour
Manually:
Looking for rdf,
owl, etc files
Harvesting vocabularies
Ratios Graphs
Reuse ratio
Detailed reuse ratios
Import graph
Reference graph
Calculating derived products (Phase 2)
Results, Analysis, and Discussion (v)
ImportGraph ReferenceGraph
• Unconnected graphs
• Few of them have in and out links
• ReferenceGraph is denser than the ImportGraph
Extracting static statistics (Phase 1)
Per element: Per ontology:
Type Name
Observed in vocabulary
Ontologies imported
Ontologies referenced
Type of appearance

14. The Landscape of Ontology Reuse in Linked Data 14
Table of contents
• Introduction
• Experimental Method
• Results, Analysis, and Discussion
• Conclusions and Future Works

15. Future
work
15The Landscape of Ontology Reuse in Linked Data
Conclusions and Future Works
In this
paper we...
• to complete the set of vocabularies analyzed so that all vocabularies appearing
in the nodes are included.
• to analyze the outliers obtained from our study as some results might be due to
o mismatches between URIs (e.g., mismatch between a URI used in an
owl:imports statement and the one use as preferred in the ontology being
imported)
o mismatches between ontology versions (e.g., the ontology retrieved when
importing a given namespace and the one found following an ontology
documentation website).
• have drawn the current reuse status in a subset of the LD vocabularies. It could
be useful for:
o Linked Data working teams aiming to reuse ontology terms
o LOV developers to include new aspects and metrics of the vocabularies in
their ecosystem
• have observed the type of appearances of elements in the analyzed
vocabularies: locally defined (59.47%), imported (27.18%), referenced (7.33%)
and referenced by import (6.02%).
• have sketched a first version of the linked vocabularies cloud overview

16. Questions
16
Thanks!
The Landscape of Ontology Reuse in Linked Data
Any questions?

17. Date: 19/10/2012
The Landscape of Ontology Reuse
in Linked Data
María Poveda, Mari Carmen Suárez-Figueroa,
Asunción Gómez-Pérez
Ontology Engineering Group. Departamento de Inteligencia Artificial.
Facultad de Informática, Universidad Politécnica de Madrid.
Campus de Montegancedo s/n.
28660 Boadilla del Monte. Madrid. Spain
{mpoveda, mcsuarez, asun}@fi.upm.es