This document discusses the development of an ontology for publishing agricultural linked data as part of the FOODIE project. It begins with context on the importance of agriculture in the EU economy. It then outlines the transformation process used to develop the ontology from the INSPIRE UML model, including pre-processing, running a tool to generate an initial ontology, and post-processing. The ontology adds new concepts like Plot and ManagementZone and defines relationships between concepts like Intervention and Treatment. Future work includes aligning the ontology with standard vocabularies and publishing linked data generated from the FOODIE data model and collected agricultural data.

1. An INSPIRE-based vocabulary for the
publication of Agricultural Linked Data
OWLED Workshop 2015
Bethlehem, PA. USA
9th October 2015
Raul Palma, Tomas Reznik, Miguel Esbri, Karel Charvat, and Cezary Mazurek
w w w . f o o d i e - p r o j e c t . e u
Grant agreement no: 621074
CIP-ICT-PSP-2013-7 Pilot Type B

2. 2www.foodie-project.eu
Presentation Outline
1. Motivation and context
2. Transformation Process
3. Ontology
4. Discussion and Future Work

3. 3www.foodie-project.eu
Fig1 - GVA
Fig2 - Employment
Fig3 – Land use
 Agriculture sector has high importance
in the EU economy
• The agriculture+ sector represented about 1,7% of
the EU-28 Gross Value Added(fig1) and accounted
for 4.9% of the total number of persons employed
in 2013 (fig2)
• Around 40 % of the total EU-28 land area is used
for farming and for forestry (fig3)
 Accordingly, EU has developed policies
and innovation programs that tackle
• the challenges associated to improve
the efficiency of agricultural activities
• with a limited environmental footprint
First some statistical facts
Source: Eurostat

4. 4www.foodie-project.eu
Agriculture sector overview
• Multiple activities and
stakeholders
• Multiple applications, tools
and devices
• Multiple data sources, data
types and data formats
In order to make economically and environmentally sound decisions, the different
stakeholders groups involved in the agricultural activities need integrated access
to multiple and heterogeneous sources of information collected by multiple
applications and devices.
Complex ecosystem!

5. 5www.foodie-project.eu
 FOODIE project aims at building an open and interoperable cloud-based platform
addressing among others the integration of data relevant to farming production
including their geo-spatial dimension, as well as their publication as Linked data.
 Data integration (and system interoperability) at the semantic level involves
• matching, aligning or translating models/schemas to provide a unified data view/access
• linking data/model elements to a reference vocabulary, e.g., AGROVOC, other well-known
 Publication of linked data involves
• dataset identification, model specification, RDF data generation and linking
• model provides an application vocabulary for representing cross-domain data and
information specific to FOODIE
FOODIE vision and goals (semantics)
High-value apps & services supporting planning and decision-making

6. 6www.foodie-project.eu
 Task1: (i) to define the application vocabulary covering the
different categories of information the platform has to deal with,
(ii) in line with existing standards and best practices
 INSPIRE directive is an EU initiative that aims at building a Pan-
European spatial data infrastructure (SDI) requiring
• EU Member States to make available spatial data, from multiple
thematic areas, according to established implementing rules using
appropriate services.
• Based on ISO/OGC standards for geographical information, i.e., ISO
19100 series standards
 Hence, FOODIE data model builds on
• the INSPIRE data specification for agricultural
and aquaculture facilities theme - AF (for
agricultural data), and
• the INSPIRE data specification for themes
in annex I for for geospatial data
FOODIE data model

7. 7www.foodie-project.eu
 INSPIRE data specifications are
defined as UML models and are
available in different XML-based
formats
 FOODIE extensions on the UML
model developed by domain experts
 Task2: Transform model into OWL
ontology
FOODIE data model
class Foodie Core Data Model v4.1 proposal
«featureType»
Agricultural and
Aquaculture Facilities
Model::Holding
«featureType»
Agricultural and Aquaculture Facilities Model::
Site
+ code :Identifier
+ geometry :GM_Object
+ activity :EconomicActivityNACEValue [1..*]
+ validFrom :DateTime
+ validTo :DateTime [0..1]
+ beginLifespanVersion :DateTime
+ endLifeSpanVersion :DateTime [0..1]
«voidable»
+ includesAnimal :FarmAnimalSpecies [0..*]
«dataType»
Agricultural and Aquaculture Facilities Model::
FarmAnimalSpecies
«voidable»
+ livestock :LivestockSpeciesValue [0..*]
+ aquaculture :AquacultureSpeciesValue [0..*]
«featureType»
Activity Complex::ActivityComplex
+ inspireId :Identifier
+ thematicId :ThematicIdentifier [0..*]
+ geometry :GM_Object
+ function :Function [1..*]
+ userId :CharacterString
«voidable»
+ name :CharacterString [0..1]
+ validFrom :DateTime
+ validTo :DateTime [0..1]
«voidable, lifeCycleInfo»
+ beginLifespanVersion :DateTime
+ endLifespanVersion :DateTime [0..1]
«featureType»
Plot
+ code :Identifier
+ validFrom :DateTime
+ validTo :DateTime [0..1]
+ beginLifeSpanVersion :DateTime
+ endLifeSpanVersion :DateTime [0..1]
+ geometry :GM_Object [1..*]
+ description :CharacterString [0..1]
+ originType :OriginTypeValue
«featureType»
Treatment
+ quantity :Measure [1..*]
+ tractorId :CharacterString [0..*]
+ machineId :CharacterString [0..*]
+ motionSpeed :Measure [0..1]
+ pressure :Measure [0..1]
+ flowAdjustment :Measure [0..1]
+ applicationWidth :Measure [0..1]
+ areaDose :DoseUnit [0..1]
+ formOfTreatment :FormOfTreatmentValue [1..*]
+ treatmentPurpose :TreatmentPurposeValue [0..*]
+ treatmentDescription :CharacterString [0..1]
«dataType»
NutrientsType
+ N :Measure [0..1]
+ P2O5 :Measure [0..1]
+ K2O :Measure [0..1]
+ MgO :Measure [0..1]
+ CaO :Measure [0..1]
+ S :Measure [0..1]
+ Zn :Measure [0..1]
+ Cu :Measure [0..1]
+ Fe :Measure [0..1]
+ B :Measure [0..1]
+ Mn :Measure [0..1]
+ Mo :Measure [0..1]
«codeList»
TreatmentPurposeValue
+ weed
+ pest
+ disease
«featureType»
CropSpecies
+ beginDate :Date
+ endDate :Date [0..1]
+ cropArea :GM_Object
+ cropSpecies :CropType [1..*]
+ production :ProductionType [0..*]
«featureType»
Product
+ productCode :CharacterString [0..*]
+ productName :CharacterString [1..*]
+ productType :CharacterString
+ productSubType :CharacterString [0..*]
+ productKind :ProductKindValue
+ description :CharacterString [0..1]
+ manufacturer :CI_ResponsibleParty [1..*]
+ nutrients :NutrientsType [0..*]
+ safetyInstructions :CharacterString [0..1]
+ storageHandling :CharacterString [0..1]
+ registrationCode :CharacterString [0..*]
+ registerUrl :URL [0..*]
«featureType»
SoilNutrients
+ nutrientName :GenericName
+ nutrientAmount :Measure
+ nutrientMeasure :CharacterString
«codeList»
ProductKindValue
+ organic
+ mineral
«dataType»
SoilTextureType
+ clay :Percent
+ silt :Percent
+ sand :Percent
«dataType»
ProductionType
+ productionDate :Date
+ variety :CharacterString
+ productionAmount :Measure
+ productionAnalysis :ProductionAnalysisType [0..*]
«dataType»
ProductionAnalysisType
+ productionAnalysisDate :Date
+ property :Measure
«codeList»
OriginTypeValue
+ manual
+ system
«featureType»
Alert
+ code :Identifier
+ type :CharacterString [1..*]
+ description :CharacterString [0..1]
+ checkedByUser :Boolean
+ alertDate :Date
+ alertGeometry :GM_Object
«featureType»
Intervention
+ type :CharacterString
+ description :CharacterString
+ notes :CharacterString [0..1]
+ status :CharacterString
+ creationDateTime :DateTime
+ interventionStart :DateTime
+ interventionEnd :DateTime [0..1]
+ interventionGeometry :GM_Object [1..*]
+ supervisor :CI_ResponsibleParty [0..1]
+ operator :CI_ResponsibleParty [0..*]
+ evidenceParty :CI_ResponsibleParty [1..*]
«featureType»
TreatmentPlan
+ treatmentPlanCode :CharacterString [0..*]
+ description :CharacterString [1..*]
+ type :CharacterString
+ campaign :TM_Period [1..*]
+ treatmentPlanCreation :DateTime
+ notes :CharacterString [0..1]
«featureType»
ProductPreparation
+ productQuantity :Measure
+ solventQuantity :Measure [0..*]
+ safetyPeriod :TM_Period
«dataType»
ActiveIngredients
+ code :CharacterString [0..1]
+ ingredientName :CharacterString
+ ingredientAmount :Measure
«codeList»
DoseUnit
+ minimumDose :Measure
+ maximumDose :Measure
«codeList»
FormOfTreatmentValue
+ manual
+ applicationMachine
+ aerial
«featureType»
ManagementZone
+ code :Identifier
+ validFrom :DateTime [0..1]
+ validTo :DateTime [0..1]
+ beginLifeSpanVersion :DateTime
+ endLifeSpanVersion :DateTime [0..1]
+ geometry :GM_Object [1..*]
+ notes :CharacterString
+ dateOfAnalysis :DateTime [0..*]
«featureType»
SoilType
+ soilType :CharacterString [1..*]
«featureType»
SoilTexture
+ soilTexture :SoilTextureType [1..*]
«featureType»
pH
+ pH :Measure [1..*]
«featureType»
OrganicMatter
+ organicMatter :Percent
«featureType»
ElectricConductivity
+ electricConductivity :Measure [1..*]
«featureType»
ZonePolygon
+ area :Measure
«dataType»
CropType
+ name :CharacterString
+ scientificName :CharacterString
+ description :CharacterString
+ notes :CharacterString
+ seedingAdvice :CharacterString
+ harvestingAdvice :CharacterString
0..*
1..*
+contains
1..*
HoldingSites
1..*
0..*
+containsZone
0..*
+HoldingPlot
0..*
+containsPlot 1..*
HoldingPlots
0..*
0..1 0..1
0..*
0..*
+crop 0..*

8. 8www.foodie-project.eu
Presentation Outline
1. Motivation and context
2. Transformation Process
3. Ontology
4. Discussion and Future Work

9. 9www.foodie-project.eu
 Semi-automatic approach* using ShapeChange tool, which
• processes application schemas for geographic information from a
UML model and derives implementation representations
• Implements ISO 19150-2 standard defining rules for mapping ISO
geographic information UML models to OWL ontologies.
• Required different pre and post processing tasks
Transformation from UML model to OWL ontology
S. Tschirner, A. Scherp, and S. Staab. Semantic access to inspire. In Terra Cognita’11 Workshop Foundations,
Technologies and Applications of the Geospatial Web. Citeseer, 2011
L. Van den Brink, P. Janssen, and W. Quak. From geo-data to linked data: Automated transformation from gml
to rdf. Linked Open Data-Pilot Linked Open Data Nederland, 2013.
XMIXML schemas, feature
catalogs, and RDF/OWL

10. 10www.foodie-project.eu
 Pre-processing tasks
• Source model preparation
 assignment of INSPIRE application schema stereotype, fixing
inconsistencies, naming target sides of aggregations/associations
• ShapeChange tool configuration
 Encoding rules driving the conversion process
 Mappings from UML classes to OWL elements
 for the classes and properties referenced in the model
 Namespaces definition for base ontologies (fixed and extended)
 ISO 19100 series standards, GeoSPARQL OGC standard and INSPIRE
 standard vocabularies like rdf, skos, dublin core and PROV
• Base ontologies fixes
 INSPIRE common ontology
 Ontologies based on the ISO 19100 series standards**
 drastic changes between versions, miss several elements from the standard
and in most cases the ontologies are only available as OWL full ontologies
Transformation from UML model to OWL ontology

11. 11www.foodie-project.eu
 Preliminary ontology generated after ShapeChange
execution
• UML featureTypes and dataTypes modelled as classes, and their
attributes as datatype or object properties
• UML codeLists modelled as classes/concepts, and their attributes as
concept members
• Cardinalities restrictions defined on properties (exactly, min, max)
• DataType properties ranges defined according to model/mappings
• Object properties ranges defined according to model/mappings
• Object properties inverseOf defined
 Post-processing tasks
• Manual fixes in the ontology
 incorrect namespaces, missing prefixes and unnecessary
imports
 incorrect axiom generated to constraint the cardinality of the
property rdfs:label in a class expression
• Manual creation of ontology elements of the base INSPIRE
schemas (AF)
Transformation from UML model to OWL ontology

12. 12www.foodie-project.eu
Presentation Outline
1. Motivation and context
2. Transformation Process
3. Ontology
4. Discussion and Future Work

13. 13www.foodie-project.eu
 For the purposes of FOODIE, we found the lack of a feature on a
more detailed level than Site that is already part of the INSPIRE
AF data model.
 Main concept: Plot
FOODIE ontology
foodie:Plot
INSPIRE-AF:Site
foodie:Alert Foodie:InterventionFoodie:CropSpecies
Foodie:ManagementZone
containsPlot
containsManagementZone
interventionPlotspeciesPlot
alertPlot
plotAlert
Foodie:ProductionType
production
 One lower level is
ManagementZone
• Enables a more precise description
of the land characteristics in fine-
grained area
• Represents a continuous area of agricultural
land with one type of crop species, cultivated
by one user in one farming mode
• Two kinds of data associated:
• metadata information
• agro-related information
Foodie:SoilNutrients
zoneNutrients

14. 14www.foodie-project.eu
 The Intervention is the basic feature type for any kind of
(farming) application with explicitly defined geometry, e.g.,
tillage or pruning.
• Has multiple indirect associations with different concepts
FOODIE ontology
Foodie:Intervention
Foodie:Treatment
Foodie:TreatmentPlan
Foodie:Product Foodie:ProductPreparation
Foodie:ActiveIngredients
is-a
plan
productPlan
planProduct
preparationProduct preparation
productTreatment
treatmentProduct
preparationPlan
ingredientProduct
Foodie:FormOfT
reatmentValue
Foodie:Treatme
ntPurposeValue
formOfTreatmenttreatmentPurpose

15. 15www.foodie-project.eu
FOODIE ontology
The Top hierarchy The FeatureType hierarchy

16. 16www.foodie-project.eu
FOODIE ontology
The Datatype hierarchy The Codelist hierarchy

17. 17www.foodie-project.eu
Presentation Outline
1. Motivation and context
2. Transformation Process
3. Ontology
4. Discussion and Future Work

18. 18www.foodie-project.eu
 Not a straightforward process, needed substantial manual
intervention
 Need to clarify the issues with the ontologies based on the ISO
19100 series standard.
 FOODIE ontology to be aligned with standard vocabularies, so that
data can be integrated with existing sources
• FAO vocabularies
 AGROVOC
 Agrontology
• Datacube (e.g., for integration with EUROSTAT)
 Source data collected and generated in compliance with FOODIE
model will be made available in a relational database
• FOODIE has a clustered postgres-XL (with postgis) installation
• UML model has been translated into a relational schema just recently
and data has started to be collected
 Publication of linked data will require data conversion (e.g., data
upgrade or on-the-fly query translation)
Discussion & Future Work

19. Partners
www.foodie-project.eu
This project has received funding from the European Union’s Seventh Framework
Programme for research, technological development and demonstration under grant
agreement no. 621074
Thank you for your attention
Raul Palma (PSNC)
rpalma@man.poznan.pl
http://www.foodie-project.eu
http://twitter.com/foodie_project