Presentation of agriopenlink @ EFITA (main program)

EFITA/WCCA/CIGR 2015, Poznan, Poland, June 29 to July 2, 2015.

EXPERIENCE WITH CREATING A PRECISION DAIRY

FARMING ONTOLOGY FROM ISOAGRINET IN

AGRIOPENLINKDana Tomic, Dr. (FTW)


agriOpenLink Key Facts

Interdisciplinary Research Project

- FTW - Research Center for ICT

- JR - Josephinum Research

- BOKU - University of Natural Sciences BOKU

- MKWE - Animal Monitoring System (SMARTBOW)

The main theme: Information Management for Processes Optimization in Precision Agriculture

Funded by: Austrian Research Funding Agency (FFG)

Use Cases: 1) Dairy Farming, 2) Irrigation

Duration: 3 Years ( 06’2013- 05’2016)

Budget: ~900 k Euro (80% funding)

2


SMARTBOW - Based on ear tags.

• Activity / resting

• Fertility

• Rumination

• etc.

- Improved nutrition and

fertility management,

animal wellbeing and

health.

80 000 per cow per day

Color codes for activities (resting, feeding, moving)


The Data Integration Problem

Advanced

analytics

Advanced

robotics

Dumping

files

User as the integrator!


Provide decision

support based on

integrated data

Plugin

agriOpenLink

The Vision

DFO

Integrate any system!

Integrate any data!

Integrate knowledge

3rd Party

Software

API for external APPs


agriOpenLink Technical Realization

6

Plugin Component

- Plugin Gateway

- Plugin Server (HTTP REST+RDF)

- Plugins (loadable components)

- Plugin Services (RDF)

Query Component

- Query Editor and EngineConsultant

Farmer

Plugin

Developer

Ontology Component

- Dairy Farming Ontology

- Repository with Ontology & Data

- DFO Editor for collaborative editing


DFO Engineering: Goals and Approaches

Phase-1 : Domain Modelling

Domain modelling by the

domain experts

Modelling in the Ontology Editor

Protege

Hierarchy of primitive classes

and, data properties and

relationships

Relating to the existing

knowledge ISOagriNet

standard)

Setting the scope and boundary

of the model (e.g., demonstrate

integration of three systems via

a number of SPARQL queries)

Goal: Include all necessary

properties classes and

relationships.

7

Phase-1 : Operational Concept Modelling

Service Ontology for the service registry

Plugin, Service, provider, operator, …, relationships

Use properties defined in the domain ontology for the

services

Goal: Support service registration and finding

Service &

Plugin

Ontology

DF

Ontology

(core)



Phase 1: Knowledge Encoding

Creating the so called “defined classes” that define the sufficient

and necessary conditions for classification

A defined class is specified with restrictions on properties: e.g.:

- A lame animal can be modeled with a restriction on the mobility score

- In heat animal can be defined with restrictions on properties from many

systems, e.g., rumination, activity, milk yield, reduced feed intake, …

Alternatively knowledge can be encoded with SPARQL queries,

e.g.

8

SELECT … (all animals with sudden

increase of resting, reduction of milking

time increase of milk conductivity,

protein, fat and lactose.) Risk of

MASTITIS!



Phase 2: Translating Existing Domain Knowledge into the Semantic Format

Triplefication = other format < subject, predicate, object>

Goal : Instead of implicit links to the standards (captured in the annotation “seeAlso” in the ontology) create explicit links to these resources modelled by using URI, RDF, RDFS and OWL schemas.

9

Exiting Domain

Knowledge but

now put in the

format


ISOagriNET Framework for Data Exchange

International standards for data exchange: ADIS/ADED standards (ISOagriNET framework)

A working group ISO ISO/TC 23/SC 19/WG 2

ADIS – the Agricultural Data Interchange Syntax - how data is transmitted

ADED – the Agricultural Data Element Dictionary - what data is transferre

10

ISO 11787 Machinery for agriculture and forestry - Data interchange between

management computer and process computers - Data interchange syntax

Since 1995

ISO 11788-1 Electronic data interchange between information systems in agriculture -

Agricultural data element dictionary - Part 1: General description

Since 1997


Agricultural data element dictionary - Part 2: Dairy farming

Since 03/98


Agricultural data element dictionary - Part 3: Pig farming

Extended 07/98

DIN 11800 Agricultural engineering - Electronic data interchange in agriculture - data

interchange in the field of cattle production

Since01/98



The German milk control association (LKV-NRW) set up a

specialized portal with information and interfaces targeting

developers who are implementing applications that integrates

exchanges based on ADIS/ADED.

– access to a database with yearly

published ADED Dictionaries

– data elements in HTML format

11

http://ian.lkv-nrw.de/index.php?id=292&no_cache=1



12

The Query Interface is a WEB GUI – for

inspection of the dictionary or download of

the dictionary in the ADIS format


ISOagriNET – General Schema

13

http://ian.lkv-nrw.de/index.php?id=292&vURL=AGRO2016%2Fadis-def%2Fe190011.html&cHash=350dc14529

Entity

Item Item Item

Key OptMan

N ANCodeSet

features

precision

length

…

Key KeyFeature

Data Dict.


ISOagriNET – Data Syntax

14

Asci

DN – lines

VN – lines…

<value, length,

precision>


Data Exchange and Interpretation based on

ADIS/ADED

15

ADED Dictionary is an

information island


ADED Data Dictionary Ont. Translation Rules

16

Entity

Item Item Item

Key OptMan

N ANCodeSet

features

precision

length

…

Key KeyFeatureADED_CodeSet

ADED_Entity

owl:Class

owl:DatatypeProperty

owl:Class

owl:NamedIndividual

Data

owl:AnnotationProperty

owl:NamedIndividual

Dict.

owl:Class


Namespaces (experimental)

Example : ADR2003

Entity

isoADR2003:Entity

isoentADR2003:884412

CodeSet

isoADR2003:Codeset

isocsADR2003:8532

Key

isocsADR2003:8532/#99

Item

isoitADR2003:DP840021

Dictionary

isoentADR2003:190001

17


Exploration od Schema

What are different entities in area “Rinder”

(beef) which contain words “Wert” and Milch in

their description (BESCHREIBUNG).

SELECT DISTINCT ?ITEM ?NAME ?BESCHR ?SPR ?TYP

WHERE { ?ITEM rdf:type owl:DatatypeProperty.

?ITEM isoitanotADR2003:BESCHREIBUNG ?BESCHR.

?ITEM isoitanot:isoarea "Rinder".

?ITEM isoitanotADR2003:TYP ?TYP.

?ITEM isoitanotADR2003:NAME ?NAME.

?ITEM isoitanotADR2003:CODEDA "N".

?ITEM isoitanotADR2003:Sprache ?SPR.

FILTER (regex(?BESCHR, "Wert", "i") && regex(?BESCHR, "Milch", "i"))}

18

OWLIM

Data

(Animal+Breed)

ADR2003

SPARQL


Exploration of Data

Exploring data that farmer receives from the control association

The result of a SPARQL query filtering the first instance of all

different data object types contained in the data file.

19


Exploration of Data

20

SELECT ?c ?g ?b

WHERE {

?c ?t ?u.

?u isoitanotADR2003:NAME “Mittlere_Laktationsleistung".

?c ?z ?b.

?z isoitanotADR2003:BESCHREIBUNG "Eiweiss-%_mittlere_305-Tage-Leistung".

?z isoitanotADR2003:BESCHREIBUNG ?g.

FILTER (?b > 3.5)

}

Filtering the

data based on

data values !


The Role of the Ontology in agriOpenLink

21


Conclusions and the Outlook

DFO is a unifying description of the domain and operational

knowledge in the dairy farming

DFO shall interlink with the external body of knowledge in the

RDF data format making it assessable for the DSS applications.

To aid to the use of already existing domain knowledge we

proposed and implemented triplefication of ADED Dictionaries.

DFO is not a static structure and requires ontology editing and

maintenance tools

The next steps:

- Data enrichment and interlinking

- A SPARQL-endpoint to this data for experimental purposes.

22



Contact

Dr. Slobodanka Dana Kathrin TomicSenior Researcher | FTW | www.ftw.atForschungszentrum Telekommunikation Wien GmbHDonau-City-Straße 1/3 | A-1220 Vienna | Austria+43/1/5052830 -54 | fax -99 | +43/6769129023

Presentation of agriopenlink @ EFITA (main program)

Engineering

domain ontology

ontology data dfo editor

data properties

service registry plugin

defined classes

necessary properties

dfo engineering

support service registration