KOS Management - The case of the Organic.Edunet Ontology

Knowledge Organization Systems

(KOS):

Management of Classification Systems

in the case of Organic.Edunet”

Vassilis Protonotarios,

Agricultural Biotechnologist, PhD

Agro-Know, Greece / University of Alcalá, Spain

Webinar@AIMS, 21/2/2014

Contents of the presentation

(Short) introduction to KOSOpen source KOS management

toolsThe MoKi toolThe Organic.Edunet ontologyUsing MoKi for managing the

Organic.Edunet ontologyNext steps

Introduction to KOS

About KOS

KOS = Knowledge Organization Systems◦a generic term used in 

knowledge organization including the following types

◦Term lists Authority Files Glossaries Dictionaries

◦Classifications & Categories Subject Headings

• Relationship Lists• Thesauri• Semantic

Networks• Topic maps• Ontologies

Focusing on ontologiesOntology: Explicit formal

specification of terms in a domain AND the relations among them 

Tree-view of an ontology

But why use KOSs?A standardized mean for referring

to the same concept using a unique name

A mean for the classification of different resources in a domain

…and of course the backbone of linking heterogeneous data sources

Open Source KOS Management tools (indicative list)

Talking about KOS managementManage entries

◦Add, revise, deleteTranslate entriesChange relationshipsImport existing lists of

terms/conceptsExport the lists as OWL/SKOS

Tools: VocBenchavailable at

http://vocbench.uniroma2.it/ developed by FAO and its partners;a web-based, multilingual, editing

and workflow tool;manages thesauri, authority lists and

glossaries using SKOS;facilitates the collaborative editing of

multilingual terminology and semantic concept information.

VocBench screenshot

Tools: ProtégéAvailable at

http://protege.stanford.edudeveloped by the Stanford Center for

Biomedical Informatics Research at the Stanford University School of Medicine;

ontology editor and knowledge-base framework;

supports modeling ontologies via a web client or a desktop client;

Protégé ontologies can be developed in a variety of formats including OWL, RDF(S), and XML Schema

Protégé screenshot

Tools: TemaTresAvailable at

http://www.vocabularyserver.coma tool for the development &

management of controlled vocabularies, thesauri, taxonomies other types of formal representations of

knowledge

ensures consistency & integrity of data and relationships between terms

TemaTres screenshot

Tools: NeologismAvailable at: http://neologism.deri.ie/ developed by DERI (Digital Enterprise

Research Institute), Irelanda vocabulary publishing platform for the

Web of Datafocuses on ease of use and compatibility

with Linked Data principles◦ facilitates the creation of RDF classes and

propertiessupports the RDFS standard, and a part of

OWLIs NOT ontology/SKOS editor and does not

support multilingual labels

Neologism screenshot

The MoKi tool

MoKi: the Enterprise Modelling

WiKiAvailable at https://moki.fbk.eu/website/index.php

Developed by FBK, ItalySupports the construction of

integrated domain & process models

Easy editing of a wiki page by means of forms

Automatic import and export in OWL and BPMN

MoKi screenshot (2011)

MoKi evolutionDuring the Organic.Lingua ICT/PSP project:

◦Multilinguality options Integration of three machine translation services

◦Ontology enrichment services Automatically suggests new concepts for the

ontology

◦Mapping component Used for mapping the OE ontology to AGROVOC

◦Collaboration options Decisions made on discussions

◦Ontology service Exposure of ontology through REST API

The Organic.Edunet ontology

The Organic.Edunet ontologya conceptual model useful for

classifying learning materials on the Organic Agriculture (OA) and Agroecology (AE) domain

Developed in the context of the Organic.Edunet eContentPlus project

Used by Organic.Edunet for the classification of educational resources

The Organic.Edunet ontologyCurrently consists of 381

concepts translated in 18 languages

Translating the OE ontology (2010)

Building the Organic.Edunet ontology (1/3)

1. OA & AE domain experts elaborated a list including all the relevant terms in the domain of OA & AE

2. Using the list of terms as input, domain experts identified sub-domains with the aim of dividing the original list into microthesauri

◦ with the help of librarians and guidance from the ontology experts

Building the Organic.Edunet ontology (2/3)

3. Domain experts added agreed, unambiguous definitions for the terms, thus producing a “concept list”

4. Ontology experts developed an initial ontology from the concept list

5. The ontology produced in the previous step was evaluated making use of upper ontologies

Building the Organic.Edunet ontology (3/3)

Evolution of the Organic.Edunet ontology using MoKi

Time for evolutionOrganic.Lingua ICT-PSP project

(2011-2014)◦Aims to enhance the multilinguality

options of the Organic.Edunet Web portal

◦provided the opportunity for updating & revising the Organic.Edunet ontology

The requirementsMultilinguality

◦Facilitate the translation processes Avoid using spreadsheets for translations Use of machine translation tools

◦Automate processCollaborative work

◦Use web-based tool◦Enable discussions for concept revisions◦Enable different translations to take place at

the same timeExposure

◦Automatic exposure of the ontology through API

The process (1/2)Formation of teams

◦Ontology experts / knowledge engineers◦Domain experts◦Language experts

Definition of tasks◦Deprecation of less-frequently used

concepts◦Refinement of most widely-used concepts◦Addition of new concepts◦Translation of concepts◦Mapping concepts to AGROVOC

The process (2/2)Development of scenarios

◦A number of scenarios was developed per task & with specific deadlines

Collaborative work◦Discussions in MoKi◦Evolution based on discussions◦Validation of revisions by experts

Discussions in MoKi

Concept managementRefers to

Editing concept Renaming concept Deleting concept

Introduction of new conceptsOntology suggestion service

Verified Keywords, User (modified) Keywords, (Automatically) Extracted Keywords and Search-Query-Logs

Translation of concepts (2013)

Mapping to AGROVOC

Exposure of conceptsOntology service = use of APIhttp://wiki.organic-lingua.eu/APIs#Ontology_Service_API

◦Publish/expose the ontology◦Enable up-to-date publishing

Two different interfaces:◦Linked Open Data (LOD): Provides

data in SKOS format◦RESTful RDF: Exposes data in OWL2

or LOD format

Case study: the use of the ontology service API

OE ontology evolution in numbers

Next steps

Next steps in the ontology evolution (1/2)

Further work on the concepts◦Introduction of new concepts◦Refinement of existing ones◦Deprecation of existing ones◦Translation of concepts in additional

languages◦Mapping of the ontology to

additional ones

Next steps in the ontology evolution (2/2)

Publication of ontology as linked data◦Definition of a namespace◦Ensure compliance with existing

standards

Link ontology with other related ones◦Already linked to AGROVOC

AcknowledgementsThe Organic.Edunet ontology was

developed in the context of the Organic.Edunet project under the eContentPlus Programme

Parts of the work described in this presentation were partially funded by the Organic.Lingua project under the ICT Policy Support Programme

www.organic-lingua.eu

Contact me at:vprot@agroknow.gr

Thank you!