1. Semantic Technologies at FAO
International Society for Knowledge Organization (ISKO)
3 Aprile 2009, Torino
Margherita Sini

2. Few words about myself

3. Just a very rapid introduction
What?
semantic, semantic web, semantic technologies
ontologies, Knowledge Organization Systems,
metadata
Why?
interoperability, exchange, share
user orientation, precision and recall
multilinguality, cultural views, context
Who?
everybody, all domains, all countries, all .org
Which instruments?
experts, NLP, methodologies and techniques

4. Outline Semantic projects involving FAO Conclusions AOS
IPFSAPH, FNA, CWR, Fisheries, Food & nutrition, Geopolitical ontology, AGROVOC Concept Server
Thai Rice Onto, Agropedia Indica
Conclusions
ontology-based applications
Collaborations
Take home messages

5. Semantic projects involving FAO

6. Inferencing Reasoning
Why AOS
Semantic navigation,
Clustering, Ranking, ...
Interoperability
ship or
container
Terminology brokering
Intelligent query
expansion
Inferencing Reasoning
Machine learning
vessel? craft? boat?
bateaux?
barco?

7. Agricultural Ontology Service
An FAO initiative for more coherence in Agricultural Information Systems
Need of a semantic approach
AOS elements:
AGROVOC Concept Server
KOS registry
Mapping registries
Metadata standards
Tools
Publications (guidelines, ...)
Built from AGROVOC
Domain concepts
Categories
AGROVOC Concept Server
Ontology registry
Sub-domain ontologies
Metadata
ontologies

8. IPFSAPH

9. IPFSAPH

10. The Ontology

11. Creation of the core ontology
Agrovoc
Information Resources
Brainstorming
Codex Alimentarius
SPS Agreement
Ontology
1600 concepts
subject specialists
Food Safety
Documents
Generic
Documents
Ontology Editor
(OI-Modeler)

12. Concept Search
The same records will be retrieved regardless of the specific synonyms or singular/plural forms that the user uses to refer to a concept.
Related concepts

13. Multilinguality
The system is also able to understand a concept even when different languages are used.

14. Check spelling
Spelling errors are corrected: e.g. “desease” into “disease”

15. Paraphrasing
“mad cow disease symptoms” or “clinical signs of bovine spongiform encephalopathy”

16. give the same results, which are ranked.

17. Semantic navigation of the bibliographical metadata (1)

18. Semantic navigation of the bibliographical metadata (2)

19. Semantic Navigation of Knowledge
parent concept(s)
children concept(s)

20. FNA

22. Creation of the core ontology
BIBLIOGRAPHIC DATABASE
CORPORATE DOCUMENT REPOSITORY
DATABASE
MERGE
RECORDS +
TRANSFORM TO RDFS
Ontology Editor
(OI-Modeler)
maintain

23. Ontology Relationships

24. The ontology concepts Publication Issue Work Subject Term Category
Article
Subject Term
Category
Author
Region
Language
Year

25. The ontology instances

26. Features Multilingual concept resolution
Get suggestions for the navigation (e.g. synonyms)
Guided query formulation
Easy navigation of the objects by following the semantic links

27. RDFa

28. CWR

29. Hierarchy

30. The project Undertaken by FAO with partners
Developed in harmony with CWR descriptor list
First version (English only) available by December 2006
About 800 core terms + acronyms + spelling variants;
Clearly definition of concepts (AGROVOC + other sources); and
Relationships: hierarchical + causative

31. More semantics Term: wild plants subclass of plants
superclass of crop wild relatives
adapted by domestication
benefits from resource conservation

32. Properties (1/2)

33. Properties (2/2)

34. Overall Solution (1/2) Subject specific Portals News feed service
Value-added
information
services
Aggregated Database View
Subject
specific Portals
News feed service
Information System (n) …
Shared layer of
interoperability
Common exchange layer (Vocabularies,Ontologies, RDF/XML)
datasetn
dataset1
dataset2
Distributed
Datasets …

35. Overall Solution (2/2)
Data

36. Fisheries

37. Fisheries Ontologies The initial goal The approach Evolution: NeOn
Making information interchangeable between ASFA, FIGIS, OneFish and AGROVOC
The approach
Creating an ontology, integrating or mapping the 3 different systems + AGROVOC
Linking of the Ontology through wrappers to the different Information Systems
Evolution: NeOn

38. FIGIS Reference Tables
ASFA
FIGIS DTD
ONE FISH
AGROVOC
Foundational Ontology
FOS core
FOS integrated
FOS merged

39. Fisheries Ontologies (2/2)
OneFish
FIGIS
AGROVOC
Aquaculture
Resource
Water
Area
land
strains
Species
life cycle
Farming
system
management
Production center
Spawning
technique
Breeding
Hatchery
Expl. form
Regulation
Farming technique
Environment
Institution
Health monitoring
diseases
suppliers
ASFA

40. Features Form versus meaning: Traditional Search Concept Search
Implemented functionalities:
synonym search
multilingual capability
terminology brokering
disambiguation
related concepts
query expansion
Basic natural language queries
Semantic navigation of bibliographical metadata
Semantic Navigation of Knowledge
Alphabetic list ...
Core Fishery Concepts ...

41. Ontology properties

42. Example
"tell me what vessels from a nearby country are currently in the marine area 50N060W within Atlantic Ocean, provided that also some Thunnus alalunga stock can be fished by those vessels, through allowed techniques"

43. Using multilingual lexicalizations
ENGLISH
SPANISH
FRENCH

44. Using hierarchically related concepts
Polyvalent Trawlers
hierarchically related concept

45. Using non-hierarchically related concepts
gears
non-hierarchically related concept

46. Help the user formulate queries
Original query: bateau de pêche
To refine your query, click on the concepts you are interested in. They will appear to the left.
Search:

47. Reconcile different vocabularies
AGROVOC or
ASFA or
other
“fishing vessels,”
“fishing boat,”
“navire de
pêche”, “fishing
vessel”,
“embarcaciones
de pesca”
AGROVOC: “fishing vessels”, “barco”, etc...
ASFA: “fishing
vessels”

48. Semantic Navigation of Knowledge: Thesaurus based
Highlighting the originator thesaurus. User can select a specific thesaurus to look for.

49. Geopolitical ontology

50. Geopolitical ontology
Incorporate geopolitical data
Will serve as a bridge to allow communication between the various systems.

51. Properties isValidFrom hasOfficialName hasCode isSuccessorOf
hasBorderWith
dependsOn

52. Nutrition Ontology

54. Procedure
<?xml version="1.0"?>
<rdf:RDF
xmlns="
xmlns:protege="
xmlns:rdf="
xmlns:xsd="
xmlns:rdfs="
xmlns:owl="
xmlns:daml="
xmlns:dc="
xml:base="
<owl:Ontology rdf:about="">
<owl:imports rdf:resource="
<owl:versionInfo rdf:datatype="
>Revision 4.0</owl:versionInfo>
<protege:defaultLanguage rdf:datatype="
>en</protege:defaultLanguage>
<rdfs:comment rdf:datatype="
>International Network of Food Data Systems (INFOODS) was established in 1984 on the basis of the recommendations of an international group convened under the auspices of the United Nations University (UNU). Its goal was to .....</rdfs:comment>
</owl:Ontology>
<owl:Class rdf:ID="c_0413">
<code rdf:datatype="
>0413</code>
<rdfs:subClassOf>
<owl:Class rdf:ID="c_041"/>
</rdfs:subClassOf>
<rdfs:label xml:lang="en">Vitamin D</rdfs:label>
</owl:Class>
=CONCATENATE("<owl:Class rdf:ID=""",J2,"""><rdfs:subClassOf><owl:Class rdf:ID=""c_",B2,"""/></rdfs:subClassOf><rdfs:label xml:lang=""en""><![CDATA[",D2,"]]></rdfs:label><code><![CDATA[",J2,"]]></code><TAGNAME><![CDATA[",J2,"]]></TAGNAME>",S2, T2,"</owl:Class>")

55. AGROVOC

56. AOS Core: the Concept Server
Other thesauri and terminologies
integration
ABACA
NT1 Food
NT2 Apple
ANIMAL
BT Organ
NT ....
Other thesauri & terminologies
ABACA
NT1 Food
NT2 Apple
ANIMAL
BT Organ
NT ....
Terminology Workbench
AGROVOC OWL
AGROVOC RDFS formats
(e.g. SKOS)
and
TagText ISO2709
mapping
Export

57. Concept Server project
Refine semantics and enrich data pool and lexicon
Develop a workbench for terminology and ontology development and maintenance.
Support information management specialists in the development, maintenance, and quality assurance of the AOS/CS
Global knowledge vs local knowledge

58. AGROVOC Concept Server
AGROVOC cleaning and refinement
Current
AGROVOC
MySQL
Improved
AGROVOC OWL
Revision
and
Refinement

59. Modelling Conversion to UTF-8 Migration to MySQL (from SQL server)
Migration to PostgreSQL (from MySQL)
Incorporated AGRIS/CARIS classification scheme (multilingual) and the mapping with AGROVOC keywords
Modified structure to store multiple classification schemes
Revised RDBMS scheme for ontology representation
Designed OWL models
Export to OWL format (v0.8a)
Export to SKOS format (v0.8a)

60. Methods Concepts from descriptors
Synonym <owl:DatatypeProperty rdf:ID="synonym">
Acronyms <owl:AnnotationProperty rdf:about="
<owl:Class rdf:about="
<rdfs:label xml:lang="en">ABA</rdfs:label>
<rdfs:label xml:lang="fr">ABA</rdfs:label>
<rdfs:label xml:lang="es">ABA</rdfs:label>
<rdfs:label xml:lang="ar">آبا</rdfs:label>
<rdfs:label xml:lang="zh">脱落酸</rdfs:label>
<synonym xml:lang="en">[8565] Abscisic acid</synonym>
<rdfs:subClassOf rdf:resource="
<rdfs:subClassOf rdf:resource="
</owl:Class>

61. SKOS SKOS export from AGROVOC Concept Server Workbench (WB)
SKOS web services
SKOS-services for DSpace plug-in
SKOS for mapping projects

62. Web Services triple store SKOS maintain AGROVOC CS Workbench export
access
Web Services
access
response

63. Ontology models (AGROVOC Concept Server, LIR, ...)
Relationships
between
concepts
Lexicalization/
Term
String
strings
terms
designated by
manifested as
Other information:
language/culture
subvocabulary/scope
audience
type, etc.
Note
annotation relationship
Relationship
All terms are created as instances of the class o_terms. All at the same level. Only one language per term.
term level
string level
concept level

64. Agropedia Indica

65. References http://www.slideshare.net/marghe_rita/1-pantnagar

66. Conclusions

67. Ontology-based applications
Better exploitation of the potentiality at the application level: powerful IR
No more words but URIs in IS
Networked Ontologies
Ontology Web services (OWS)

68. Collaborations With AOS partners Within EU Projects
NeOn
SEMIC.EU
With other initiatives
GFIS
Ecoterm
Mapping projects
GBIF Global Biodiversity Information Facility secretariat
JRC + BGS + Biblioteca Nazionale di Firenze

69. Take-home message
There are many uses for terminology & ontology systems in food and agriculture, both for information access and information processing
FAO has several projects using such systems
FAO is deploying the Agricultural Ontology Server (AOS) as a global resource
SKOS and other knowledge representation standards play a key role

70. Questions? Thanks Margherita Sini: margherita.Sini@fao.org
Johannes Keizer:
Dagobert Soergel:
Asanee Kawtrakul:
But Also: Gudrun Johannsen, Boris Lauser, Claudio Baldassarre, Gauri Salokhe, Marta Iglesias, Caterina Caracciolo, Sachit Rajbhandari, Jeetendra Singh, Mary Redahan, Shrestha, Prashanta, Ton, Imm, Thanapth, Trakul, and many others...