• Like


Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

10 years Agricultural Ontology Initiative: Building Blocks for a Linked Data Infrastructure

Uploaded on


More in: Education
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads


Total Views
On Slideshare
From Embeds
Number of Embeds



Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

    No notes for slide
  • Some auto appreciation
  • This is the AGROVOC SKOS model that has been developed and decided in April 2010 under active collaboration from Tom Baker, who was member of the W3C SKOS working group.
  • AgriOcean Dspace release: 09/2010 Setting up of AgriOcean Dspace Community on AIMS: 09-10/2010 exchange of experiences Pilot Testing: “Bangladesh Agricultural Universities’ Institutional Repository” Cooperation with DURASPACE Informing about the new AgriOcean Dspace implementations Collaborating in the Dspace Ambassadors Program
  • This is a snapshot one year later. The growth is enormous. A central point is DBPedia, “triplified” information from Wikipedia. The different colours represent the different information types, being “life sciences” and “publications” the most populated areas, but with the area “government” strongly growing Interesting newcomers in the last months are the two VIVO datasets from the United States descriping expertise in Science. Vivo is actually a project that started the agricultural library of Cornell University
  • In a bibliographical record there is much more hidden information than displayed with the metadata. Many of the highly structured data are linking to other information on the web. In AGRIS we have now introduced something what we call “naive linking”. An AGRIS record links automatically to Google Maps for the location of the center and to Google to retrieve the full text of the resource, citation lists or other publications from the authors. This often works, but clearly not alway, s as it is not controlled by semantics, but only through identy of strings. For an uneducated machine unfortunately COW and C.O.W. are the same, whereas peanuts and groundnuts are something different.
  • The table shows 3 descriptors that are in AGROVOC, EUROVOC and UNBIS. In AGROVOC and EUROVOC they are already encoded as URIs. Easily we could establish relationships like owl.sameAs between the concepts or skos:exactMatch between labels.
  • If resources are marked up with semantically defined and machine readable concepts, they can be linked and mashed up precisely as we have seen in the example from the BBC. In this example we start with an AGRIS record on Hazardous waste, which is indexed with AGROVOC. Already now we can easily link to material indexed with Eurovoc, here an example from EuroLex. If the UNBIS thesaurus would be restructured to a concept scheme and published as LOD, related UN documents could be attached automatically by the machine.
  • How does this work: A resource is connected with each concept URI in the web. The concepts between three vocabularies are having same literal which is connected with owl:sameAS/exactMatch relationship. As we are speaking about thesauri and not ontologies we kept the relation to be chosen purposely vague. The concepts could be matched with owl:sameAS or the terms could be matches with SKOS:exactMatch. A lot of discussion on this is ongoing
  • One of the groundbreaking enterprises in this area is Thomson Reuters “Open Calais”. This is a webservice that provides semantic mark up for any unstructured text that you feed into their service The service is free of Charge. Why? I will show you later.
  • My team in collaboration with the Indian Institute of Technology in Kanpur is developing a similar service for our subject area.
  • We have here a text from 1964 without a bibliographic record at hand about a plant protection issue
  • Open Calais is very good in those areas, in which they have their own elaborated conceptscheme against which the texts are analyzed: “Places”, “Persons”, “Business Processes” , “Industry Terms”, but it is weak in the specific topic analysis, what they call “social tags”
  • AgroTagger still lacks many of the sophisticated features of “Open Calais” , but is much, much better in the subject analysis of the text
  • We will now try a life demo


  • 1. 10 years Agricultural Ontology Initiative: Building Blocks for a Linked Data Infrastructure
    • Dr. Johannes Keizer
    • FAO of the United Nations
    • Office of Knowledge Exchange, Research and Extension
    • Team Leader “Knowledge Standards and Services”
  • 2. The Internet!
  • 3.  
  • 4. Aggregation States of Knowledge
  • 5. Data Flows and Repositories in Research
  • 6.
    • “ ... FAO’s principle task is to work to ensure that the world’s knowledge of food and agriculture is available to those who need it when they need it and in a form which they can access and use ...”
  • 7. AOS Vision in 2001
  • 8. ..from thesaurus to Ontologies….
  • 9.
    • Our push of AGROVOC to the Semantic Web had enormous positive effects, among others
          • From 4 to 20 language versions
          • Defacto standard for indexing in many areas
          • More than 2000 downloads only in 2009
          • SKOS incorporated all our requirements
    • For many purposes we need semantics on a lower level than of a fully elaborated ontology – but we need them urgently
    • The development of specific Ontologies should be always application driven – a demand economy
    Lessons Learned
  • 10. AOS - Today Semantics Tools Linked Data Community
  • 11.
    • Community
  • 12. The AOS Community
  • 13. http://aims.fao.org/community/home
  • 14.
    • Better Semantics
  • 15.
    • around 30,000 concepts
    • 600000 labels in around 20 languages.
    • one-stop shop for terminological knowledge related to agriculture in general
    • a knowledge base of related concepts organized in ontological relationships (hierarchical, associative, equivalence) ‏
    • Is a concept/term/string based system
    • Concepts may be organized in multiple categories .
    AGROVOC concept scheme
  • 16. The AGROVOC concept scheme skos:broader 8171 1474 skosxl:altLabel skosxl:prefLabel skos:broader has_synonym SKOS Label SKOS Concept rdf:type rdf:type 6211 skos:broader AGROVOC Concept Scheme skos:topConceptOf skos:inScheme Other scheme in FAO skos:inScheme 12332 Further schemes in FAO :bar has_synonym has_translation skos:literalForm “ maize” :foo maïs (fr) : foo has_synonym skos:literalForm “ corn ” :bar Another scheme in FAO
  • 17. FAO FRBR Model & Authority Data Work Expression Manifestation Item Subjects Corporate Bodies Conferences Journals Series FAO Projects
  • 18. EXAMPLE: JOURNAL CONTENT MODEL isSpatiallyIncludedIn isPublishedBy isOtherLanguageEditionOf isFollowedBy/Follows
  • 19. A Model to create Linked Data
    • Rich set of relationships
    • Easier implementation of concept-based thesauri and authority data
  • 20. Geopolitical Ontology OEKM FAO of the UN
  • 21. Fishery Ontologies
  • 22. What Partners have done…..
    • Rice Knowledge Models
    • Rice Production Ontology
    • ASFA Thesaurus
    • VIVO Ontologies
    • Look to the many presentations on this workshop
  • 23.
    • Tools
  • 24. The Concept Scheme Work Bench
  • 25. Drupal AgriDrupal is a “suite of solutions” for agricultural information management and dissemination, built on the Drupal CMS by different Institutions and individuals who are now sharing their experiences in the AgriDrupal community
    • Drupal, a semantic web enabled CMS
        • General purpose CMS necessary
        • Drupal has flexibility to manage all information types
        • Drupal has a strong user community
        • Version 7 natively with RDF backbone, but semantic applications already possible with version 6
        • Drupal can be a producer and consumer of Linked Data
  • 26. Drupal Query run on a Drupal website from a Virtuoso test environment at http:// demo.openlinksw.com/sparql_demo / SPARQL endpoint RDF triples Drupal website
  • 27.
      • Objectives:
        • Assure quality in metadata creation
        • Sharing information in a standardized manner
        • Use of common semantics and interoperable syntaxes
        • Use of more sophisticated and specialized metadata
        • Use of controlled, multilingual vocabularies
      • Requirements:
        • AGRIS AP compliancy
        • AGROVOC
    AgriOcean Dspace – a tool for Repositories
  • 28. AgriOceanDspace – Thesaurus plug in
    • Developed by Kasetsart University (Bangkok, Thailand)
    • Thesaurus plug-in
        • Web services: use local or remote version of AGROVOC thesaurus/SKOS
  • 29.
    • Linking Data
  • 30. The Linked Data Universe: http:// www.linkeddata.org (july 2010)
  • 31. http://agris.fao.org/agris-search/search/display.do?f =2004/ZA/ZA04002.xml;ZA2004000049
  • 32. Linking vocabularies AGROVOC EUROVOC UNBIS Relationship http://aims.fao.org/aos/agrovoc/c_207 http://eurovoc.europa.eu/219055 agroforestry skos:exactMatch/ owl:sameAs http://aims.fao.org/aos/agrovoc/c_4826 http://eurovoc.europa.eu/220018 MILK skos:exactMatch/ owl:sameAs http://aims.fao.org/aos/agrovoc/c_12332 http://eurovoc.europa.eu/219871 MAIZE skos:exactMatch/ owl:sameAs
  • 33. http://aims.fao.org/aos/agrovoc/c_7825 http://eurovoc.europa.eu/218754
  • 34. Linking data through common URIs
    • skosxl: literalForm
    http://eurovoc.europa.eu/219871 Maize skosxl: literalForm Maize http://aims.fao.org/aos/agrovoc/c_12332 AGROVOC skosxl: literalForm Maize http://aims.fao.org/aos/agrovoc/c_12332 owl:sameAs http://eurovoc.europa.eu/219871 owl:sameAs/exactMatch http://agris.fao.org/agris-search/search/display.do?f=1996/TR/TR96001.xml;TR9600026 owl:sameAs/exactMatch http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2010:202:0011:0015:EN:PDF http://unbisnet.un.org:8080/ipac20/ipac.jsp?session=128F308557F34.283092&profile=bib&uri=full=3100001~!685149~!1&ri=1&aspect=subtab124&menu=search&source =~!horizon Maize Eurovoc UNBIS
  • 35. What are we doing with unstructured data?
    • We have enormous amounts of unstructured material
    • Still most of the documents that we are producing are mostly semantically unstructured
    • Human work to catalogue and index is becoming always more rare
    • We need machines to do automatic semantic mark ups of text
    • If machines are trained and based on concept schemes, ther are able to do so
  • 36.  
  • 37.
    • Does Concept identification in unstructured texts
    • Uses Agrovoc as a controlled vocabulary
    • Prototype under testing with excellent results (entire repository of ICARDA indexed)
    • Will produce in future Structured RDF files that can be used to link data like “open Calais ”
  • 38.  
  • 39.  
  • 40.  
  • 41. Life Demo: Semantic mark ups: http:// viewer.opencalais.com / http://agropedialabs.iitk.ac.in/Tagger/Agrotagger_text.php
  • 42. The CIARD RING
  • 43. AGRIS Linked Data
  • 44. Thank You! http:// www.ciard.net http:// ring.ciard.net http:// aims.fao.org http://agris.fao.org