Use of Semantic Artefacts in Agricultural Data-Driven Service

Development

Silke Cuno

and Philipp L

ammel

Fraunhofer Institute for Open Communications Systems FOKUS, Kaiserin-Augusta-Allee 31, 10589 Berlin, Germany

Keywords:

Data Sharing, Agricultural Digital Integration Platforms, Semantic Artefacts, Agri-Food, Ontologies Decision

Support, Interoperability, Digital Agriculture, Smart Farming.

Abstract:

The paper is a survey of the resources and efforts in the ﬁeld of semantic interoperability in the agricultural

domain, and describes challenges and solutions for building federated digital agricultural integration platforms

that provide service components based on shared and reused data. It also shows the state of the art of public

semantic artefacts and their potential contribution to solving semantic interoperability within digital agricul-

tural integration platforms by combining their main standards into common agricultural information models

that can be used by developers and ﬁt a wide range of agricultural use cases.

1 INTRODUCTION

January 2024 is marked by Europe-wide protests by

farmers against bureaucratisation, subsidy cuts, en-

vironmental regulations and more. The reasons are

many. Farmers say they are being asked to do more

for less (Deutschlandfunk, 2024; FAO, 2021). At the

ﬁrst Agri-Digital Conference in Brussels in December

2023 (EC, 2023a), it was discussed that farmers’ bu-

reaucratic burdens, worries about yields, incomes and

workloads could be reduced through the widespread

use of digital decision support systems. For example,

they could help farmers with speciﬁc crop, resource

and production planning. Monitoring systems could

detect early signs of disease or pests, allowing timely

action to be taken and signiﬁcant crop losses to be

avoided. For rural areas in particular, digital services,

ideally based on IT ecosystems, hold the promise of

improving quality of life, ensuring geographical bal-

ance, food security, spatially balanced development,

economic prosperity and the achievement of sustain-

ability goals. An IT infrastructure shared by farmers

and government agencies, providing embedded ser-

vices based on an “agricultural data space”, could be

one solution, as both development agencies and farm-

ers need an integrated view of the latest agricultural,

economic and environmental data for analysis to fa-

cilitate their work and, for example, to maintain food

https://orcid.org/0000-0003-4796-7326

https://orcid.org/0000-0002-4411-0557

security (EC, 2023e). There is a lack of decision sup-

port and reasoning systems based on integrated infras-

tructures of public data sources when it comes to plan-

ning to assess yield gains or what foods might become

scarce. At the same time, the landscape of siloed agri-

food IT services and systems is expanding. Farmers

are increasingly using Farm Management Information

Systems (FMIS), but the services are often isolated

systems. The data produced is mostly stored locally

or in closed vendor clouds, and is difﬁcult to ﬁnd and

access. Solutions remain proprietary. For the EU,

this hinders the development of the European Data

Economy. To mitigate this, the EU Data Strategy pro-

motes common data infrastructures with data spaces

for data reuse and sharing to enable the rapid creation

or integration of new innovative services. The EU’s

“Data Act 2023” legislation (EU, 2023) aims to: 1)

ensure a fair distribution of the value of data among

actors in the digital environment, 2) stimulate a com-

petitive data market, 3) open opportunities for data-

driven innovation, and 4) make data more accessible

to all (EC, 2023b). These actions are supported by

funding policies for the creation and sharing of ﬁnd-

able, accessible, interoperable and reusable (“FAIR”)

data (Wilkinson et al., 2016; Jonquet et al., 2023).

EU research initiatives and projects have produced

successive results in recent years. Progress has been

made towards solutions for ”technical interoperabil-

ity” through ”reference architectures” and data-driven

”data-space concepts” involving all sectors. How-

ever, data integration in agriculture still suffers from

346

Cuno, S. and Lämmel, P.

Use of Semantic Artefacts in Agricultural Data-Driven Service Development.

DOI: 10.5220/0012760400003756

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 13th International Conference on Data Science, Technology and Applications (DATA 2024), pages 346-357

ISBN: 978-989-758-707-8; ISSN: 2184-285X

numerous incompatible format speciﬁcations and en-

coding systems, as well as informal and undeﬁned se-

mantics at the data level (Subirats-Coll et al., 2022).

In this domain, data integration is highly intercon-

nected with other sciences such as meteorology, geog-

raphy, space, biology, etc. Domain standards are typi-

cally not-interoperable in terms of semantics and syn-

tax. And agriculture is producing increasing amounts

of real-time raw data from multiple sources, such as

soil sensors, drones, etc., which also need to be inte-

grated. The question is how best to manage data the

reuse and sharing of data in terms of “semantic in-

teroperability” for all the different use cases. Can a

“common information model” be the game changer

for digital agricultural integration platforms? In this

context, the numerous semantic artefacts and ontolo-

gies that have been developed over the last decades

are once again the focus of interest in computer sci-

ence research. To what extent can their knowledge,

methods and standards support seamless data integra-

tion? Their study and reorganisation is experiencing

a second renaissance in the age of data-driven service

development and behavioural AI (Bochtis et al., 2022;

Zeginis et al., 2023). Originally developed in the hu-

manities, semantic artefacts were ﬁrst used in IT af-

ter the turn of the century in connection with the Se-

mantic Web and Linked Data, as reasoning tools for

AI (Hoekstra, 2009). Now, in the context of agricul-

tural digital integration platforms, it is being explored

how they can be useful as components in federated

service architectures for standardised data integration

and, more recently, how they work in combination

with AI technologies (Subramanian et al., 2024).

This paper discusses whether the content and stan-

dards of semantic artefacts, fed into standardised

components of digital agricultural integration plat-

forms as information models or similar, could be a

way forward to facilitate service development and

data reuse. A survey of semantic interoperability re-

sources and efforts in the agricultural domain is pre-

sented in this paper. The paper is structured as fol-

lows: Section 2 presents related work in service ar-

chitecture, Section 3 presents the initiatives on seman-

tic artefacts and selected research projects integrating

information models for service development, and ﬁ-

nally Section 4 summarises the main ﬁndings.

2 THE DATA INTEGRATION

CHALLENGE

2.1 Interoperability Aspects

The sharing and re-use of data on federated ser-

vice platforms requires “interoperability”. This is

a critical enabler for interacting towards mutually

beneﬁcial goals, involving the sharing of informa-

tion and knowledge between organisations, through

the business processes they support, by exchanging

data between their ICT systems (EC, DG Informatics,

2017). The “New European Interoperability Frame-

work Model” (EIF) speciﬁes six interoperability lay-

ers. This paper addresses semantic and technical in-

teroperability. “Technical Interoperability” covers the

applications and infrastructures that connect systems

and services. Aspects include interface speciﬁca-

tions, interconnection services, data integration ser-

vices, etc. (EC, DG Informatics, 2017). “Semantic

Interoperability” means that the precise format and

meaning of exchanged data and information is pre-

served and understood throughout the exchange be-

tween the parties, in other words ”what is sent is what

is understood” (EC, DG Informatics, 2017). It in-

volves the deﬁnition of data structures and elements to

describe the exchange of data. One of the underlying

keys to semantic interoperability is syntactic interop-

erability, which describes the ability to exchange data

between systems (Bochtis et al., 2022). Interoperabil-

ity issues are not unique to the agricultural sector.

2.2 Developers’ Perspective

IT developers tasked with building integrated solu-

tions or reusing and sharing data are faced with an

uncertain landscape of systems, formats, data sources

and standards. Problems arise in interpreting the

meaning of this data at a semantic and syntactic level.

Documentation is often scarce or, in most cases, not

aligned with common data models. Developers are

challenged with conceptualising different syntaxes

for identiﬁers, ﬁnding, interpreting and accessing dig-

ital data sources in databases or repositories. Much

time is spent searching for the meaning in the con-

text of the application being programmed and organ-

ising data from databases. Developers need in-depth

knowledge of the structure of digital data sets and do-

main knowledge of the context. Having found the

data for the relevant use case, the usual way to de-

velop an IT service is to standardise the formats, data

models and vocabularies of the data used between the

concrete exchanging parties of the concrete service to

be developed. This requires the exchanging parties

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

347

to agree on the interfaces in advance, and this pro-

cess usually leads to a lengthy standardisation process

that is set up only for that speciﬁc service. Typically,

the developer designs the underlying data or informa-

tion model from scratch. A second way of develop-

ing services, data reuse and sharing, where semantic

artefacts and knowledge repositories provide the con-

cepts, emerged around 2010. This method promises

to be more effective. The assumption is that if the se-

mantic resources (such as ontologies) based on linked

data standards are available online, service develop-

ers will be able to look up the form and meaning of

the data they receive at runtime and react accordingly.

This second option has been implemented in early re-

search projects such as the “iGreen Project” (DFKI,

2014). The advantage is that developers don’t have

to develop a separate data format or data model for

their service, but can take depending on their acitivity,

for example, either an instance of the reference ontol-

ogy (e.g. a knowledge graph) or an ontology as in-

put. This would come in a JSON standard for linking

data (JSON-LD) for the service to be developed. The

beneﬁts are clearly seen in the reduced development

effort. The disadvantages of this approach are that a)

developers are not trained in the use and combination

of semantic artefacts and b) understanding the ontolo-

gies, semantic resources, data formats, their interfaces

and dependencies may require time-consuming anal-

ysis using speciﬁc development tools and skills. De-

velopers still need to spend a considerable amount of

time studying scientiﬁc publications, standard spec-

iﬁcations or learning from the ontologies and their

documentation in order to reuse data. Although not

directly related to the developers, another major dis-

advantage is that the adaptation workload - which can

be quite extensive - is essentially shifted to the data

providers meaning that they have to encode their own

data in a way that is usable in the context of the Se-

mantic Web (e.g. Web Ontology Language (OWL)).

2.3 Reference Architectures

Reference Architectures (RAs) are conceptual frame-

works that play an essential role in building ICT in-

frastructures. Their design principles ensure secure,

reliable and interoperable solutions. They describe

a standardised architecture that provides a reference

framework for vertical domains, such as agriculture,

mobility or related industries. An RA provides a com-

mon understanding of processes, data structures and

the underlying technologies in general. They provide

a standardised blueprint for the design and implemen-

tation of ICT systems within speciﬁc domains. Most

importantly, they provide guidelines for the integra-

tion of different technologies, data sources and stake-

holders, facilitating the seamless ﬂow of informa-

tion and enabling efﬁcient decision-making processes

(LeanIX, 2023). The development of RAs began in

the nineties in the telecommunications domain. The

design principles were service orientation (SOA), ob-

ject orientation, distribution, decoupling of software

components and separation of concerns to ensure in-

teroperability, portability and reusability of compo-

nents. Technologies remain independent and the sys-

tem is manageable by different stakeholders. Aspects

such as security, ﬂexibility and personalisation were

built into RAs. They have subsequently been adopted

in many sectors. First, the “IDABC” (EC, 2024)

adopted the framework around 2005. The aim was to

enable the public sector to deliver “interoperable ser-

vices” across Europe. The use of RAs as a conceptual

framework then became commonplace. The roll-out

to other domains started around 2010. First versions

of RAs for “Smart Agriculture” have been available

since 2011 through the EU projects SmartAgriFood

(SmartAgriFood, 2015) and FIspace (FIspace, 2015),

later FIWARE (FIWARE, 2024). They developed a

basic design for an agricultural RA (SmartAgriFood,

2015; FIspace, 2015) and brought together various so-

lutions, information systems and networked devices

in agriculture (FIWARE, 2018). FIWARE was then

transferred to the FIWARE Foundation, which has

continued developments around service integration,

produced many standards and played the role of a

major stakeholder. FIWARE RA has undergone con-

tinuous technical development. It has applied its RA

developments to different sectors such as health and,

since 2016, agriculture. In 2019, FIWARE deﬁned the

“smart agri-food domain” as a focus area, along with

smart cities, smart energy, etc. (FIWARE, 2018). The

“NGSI-LD” standard is an example of a result compo-

nent of the FIWARE initiative that is considered to be

of current relevance in the area of digital integration

platforms for agriculture in the current Demeter Agri-

cultural Information Model project (Rodriguez et al.,

2018; SmartAgriFood, 2015; Bochtis et al., 2022).

2.4 Data Spaces

RAs laid the foundation for service-oriented techni-

cal interoperability. Around 2017, new challenges

emerged with data-intensive AI and the need for in-

teroperable data layer concepts - the proposed solu-

tion was Data Spaces. The term Data Space refers

to a type of data relationship between trusted partners

that agree on common governance rules for data stor-

age and sharing within one or more vertical ecosys-

tems (EC, 2020d). Data Spaces are about federated

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

348

data sharing and reuse within federated interoperable

services, ensuring so-called data sovereignty. Sev-

eral Data Space initiatives were launched and IDSA

(EC, 2020c), FIWARE, BDVA/DAIRO and GAIA-X

formed an alliance. Their results became a strate-

gic part of the “European Data Strategy” (EC, 2020a)

2020, which aimed to create a single “European Data

Space”. The ﬁrst use case of a Data Space is to ensure

data sharing across supply chains - in the case of agri-

food, from “farm to fork”. One of the most important

aspects of the Data Space is that the data is not stored

centrally, but at the source, with the data provider.

Data is held exclusively by the members of the fed-

eration. Data Spaces are developed on a domain ba-

sis, but can be activated across domains. The Gaia-

X framework enables the creation of a sovereign and

federated agricultural data space that can be deployed

securely and at scale. Through the data space concept,

companies should have easy access to an almost inﬁ-

nite amount of high quality industrial data, creating

value while minimising the environmental footprint.

Data sovereignty and trust between members are es-

sential for Data Spaces.

Under the umbrella of the European Single Data

Space, the EC is funding the implementation of se-

lected domain-speciﬁc data spaces, such as an “Agri-

cultural Dataspace” (EC, 2020a), to improve the pro-

cessing and analysis of related data, enabling precise

and tailored application of production approaches at

farm level. In describing the evolution of IT in agri-

food (Wolfert et al., 2023), Wolfert et al. note that

the nature of digitalisation has become more com-

plex along two axes: Figure 1 shows that the level

of IT integration on the x-axis has evolved from in-

dividual applications to a complex “system of sys-

tems.” On the y-axis, the number of actors involved

increases from individual process participants to com-

plex business ecosystems with a large number of ac-

tors. This also expands the scope of RAs and Data

Spaces: 1) from the former production process to

the supply chain to the whole food system and data

economy and 2) from individual applications to oper-

ational and chain information systems to Data Spaces

as part of the “European Data Strategy” (EC, 2020a).

There are Data Space projects for agriculture on-

going: Germany is funding two called AgriGaia

(AgriGaia, 2023) and NaLamKI (NaLamKI, 2023)

and at EU level we have “AgriDataSpace” (Agri-

DataSpace, 2024). The aim is to map the current land-

scape of ongoing data sharing initiatives in agricul-

ture, to design approaches based on Data Space con-

cepts, to explore appropriate solutions for semantic

interoperability, and to analyse and evaluate current

governance models. AgriDataSpace is developing a

Figure 1: Evolution of IT in Agri-Food. Note: The el-

lipse indicates the current state, in which innovation ecosys-

tems have become complex. Adapted from Wolfert et al.

(Wolfert et al., 2023) based on (Wolfert et al., 2021).

conceptual RA with an “Information Model” (IM)

to represent concepts and relationships, constraints,

rules and operations to specify data semantics for the

domain. In general, an IM should provide a shareable,

stable and organised structure of information require-

ments or knowledge for the domain context. As a

component, an IM should be open and technology and

domain agnostic. It should enable the comprehensive

description of data assets and the semantic interoper-

ability required for this type of exchange. For an IM,

the semantic interoperability of the domain needs to

be based on standards so that data can be integrated

for further decision support. A number of European

projects using RAs for data exchange have indeed im-

plemented such an IM for agriculture on a demonstra-

tor basis using different approaches, as presented in

section 3.3 of this paper. They haven’t yet produced a

standard IM for agriculture, but research is moving in

that direction.

3 SEMANTICS FOR

INTEROPERABILITY

An important motivation for the design of data spaces

in agriculture is to support the efﬁcient development

of services for the entire food supply chain activi-

ties, ranging from agricultural production to distri-

bution and consumption, while enabling analysis and

reporting tools for health, food security and sustain-

ability. Clearly, developers know the requirements

and expected outcomes of the service they are devel-

oping. But beyond that, when it comes to providing

openness and interoperability in the service to be im-

plemented within a broader architecture, as required

by the “European Data Strategy” (EC, 2020a), IT

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

349

developers need to have a cross-domain knowledge,

namely of core information models and APIs, so that

their service can be used across application domains

such as Smart Cities, Smart Industry, Smart Agricul-

ture, and additionally they also need a deeper under-

standing of the terms in the speciﬁc domain and use

cases in order to efﬁciently design or select the appro-

priate conceptual scheme for the required data model.

As this requires a high level of expertise in domain

concepts as well as cross-domain and core standards,

and as such expertise appears to be rare, a standard-

ised information model for the domain could make

the process more efﬁcient on the service development

side. Without supporting standards, building the data

model from scratch appears to be time-consuming,

as the different concepts that may exist for a given

term, depending on the semantic artefact in which it

is found, need to be carefully investigated. For ex-

ample, (Baker et al., 2019) shows a side-by-side visu-

alisation of the data (properties and values) provided

by the semantic artefacts GACS (GACS, 2024) and

FoodOn (FoodOn, 2024), proposing comparable but

not identical concepts for maize. The choice of an

appropriate concept based on terms is crucial to en-

sure that the code is not only syntactically correct, but

also logically meaningful and functional. Develop-

ers typically capture this understanding in “Informa-

tion Models”, from which a program can perform au-

tomated reasoning. Such models can also be called

“knowledge representations” of the domain (Hoek-

stra, 2009) and are derived from semantic artefacts.

Semantic artefacts in general use speciﬁc “semantic

technologies” to describe the knowledge of a domain.

They cover the perspectives, concepts and terminol-

ogy of their provider’s domain of interest, which may

vary widely even within the same domain. They re-

ﬂect “concepts” as building blocks of a speciﬁc do-

main theory. Semantic artefacts include “thesauri”,

“taxonomies”, “controlled vocabularies” and “ontolo-

gies”. Many have in common that they are based on

research data. (Subirats-Coll et al., 2022) highlights

two basic principles for using semantic resources to

achieve semantic interoperability in an IT service:

This is the use of a globally unique web identiﬁca-

tion - a “URI” (Uniform Resource Identiﬁer) - to iden-

tify a logical/physical resource used by web technolo-

gies, and secondly, a data model that can accommo-

date all existing data sets and formats, essentially a

graph-oriented data model. The “Resource Descrip-

tion Framework” (RDF) follows these principles of

linked data and is seen as a catalyst for semantic inter-

operability (Subirats-Coll et al., 2022). The “Linked

Data Concept” (Berners-Lee, 2006) is based on this,

dealing with redundancy and handling it with links.

Semantic resources can appear in different knowl-

edge representation languages such as OWL, RDFS,

SKOS, OBO and UMLS-RRf etc.

(Baker et al., 2019) arranges semantic artefacts

or resources on a “semantic spectrum” ranging from

“lightweight” to heavyweight”. Leightweight re-

sources include e.g. “controlled lists” as standard-

ised and organised arrangements of words or phrases

or as thesauri and taxonomies with a hierarchical

structure. Languages such as RDF/RDFS, SKOS,

UML are suitable for representing these, as no rea-

soning is required (Hoekstra, 2009). In the “semantic

spectrum” given by (Baker et al., 2019), the seman-

tic resources GACS and ARGROVOC (FAO, 1981)

are labeled as “concept schemes” and classiﬁed as

“middleweight”, whereas the well-known FoodOn,

Crop (Ontology, 2024a) and Gene Ontology (Ontol-

ogy, 2024b) are labelled as “Ontologies” and “heavy-

weight”. They represent knowledge and are reusable

terminological knowledge representations (Hoekstra,

2009). Ontologies are semantically described with

rich metadata (Drury et al., 2019). Languages suitable

for representing ontologies are developed to support

ontology-based reasoning services within knowledge-

based applications. The main standards used are the

OWL-DL family (Web Ontology Language Descrip-

tion Logic) (Hoekstra, 2009; Baker et al., 2019). Over

the last few decades, international organisations have

built up large semantic artefacts for agronomy and

have carried out a multifaceted collaboration on ed-

itorial service tools, concept mapping, etc. Examples

include the prominent AGROVOC of the Food and

Agriculture Organisation (FAO), the NAL Thesaurus

(NALT) (NALT, 2024) of the National Agricultural

Library (NAL) and the CAB Thesaurus of the Cen-

tre for Agricultural Bioscience International (CABI,

2024). Developers can ﬁnd them in repositories such

as AgroPortal (AgroPortal, 2024) or Fairsharing.org

etc. In addition to these general agronomic semantic

resources, there is a wide range of speciﬁc semantic

resources, all of which provide speciﬁc exchange for-

mats and associated services. A good overview can

be found in (Arnaud et al., 2020). Nevertheless, the

use of ontologies in IT service development remains

underexploited (Drury et al., 2019). As argued above,

one reason may be that the proliferation of semantic

resources that redeﬁne the same concepts, even when

based on linked data, is not straightforward for the av-

erage developer. Key questions are on how to improve

the linking of various resources (Subirats-Coll et al.,

2022) and how to encourage developers to use this

complex and rich knowledge of semantic artefacts for

service development. The next two sections describe

some domain-speciﬁc semantic artefacts and reposi-

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

350

tories that provide implementations of Semantic Web

technology, rich concepts and datasets, and SPARQL

endpoints that may be relevant to developers.

3.1 Selected Semantic Artefacts

AGROVOC. (FAO, 1981; Caracciolo et al., 2013)

is known as the world’s most widely used RDF-based

semantic resource for agriculture. It covers the topics

food, nutrition, agriculture, forestry, ﬁsheries, envi-

ronment, etc. It is hierarchically organised under 25

top concepts with deﬁnitions and relationships. It is a

multilingual “controlled vocabulary”. It was created

in 1981 and has since then been continuously devel-

oped and maintained by a network of institutes. Tech-

nically, it has moved from databases to the Seman-

tic Web with Linked Open Data (Caracciolo et al.,

2013). It collaborates with other knowledge resources

and is constantly expanding its coverage (Brewster

et al., 2023; Baker et al., 2019). It is edited using the

web-based platform VocBench, a software for man-

aging controlled vocabularies such as ontologies, the-

sauri and generic RDF datasets (Stellato et al., 2020).

AGROVOC can be accessed, searched for concepts

or browsed by hierarchy. It can be downloaded as

an RDF dataset. Available web services can be used,

searches can be performed via SPARQL queries, us-

ing public SPARQL endpoints and as an interface for

submitting RDF queries. Recently, AGROVOC has

been brought into the context of a “Data Space for

Agriculture” in (D

orr et al., 2023). Data Spaces can

be seen as data sharing spaces, but AGROVOC, even

if it is based on linked data, would be a ”vocabulary”

component within a Data Space. And a data space is

an organised community of data providers and data

users who contractually agree to share their data ac-

cording to rules, and whose members have commer-

cial objectives. In contrast, AGROVOC is open to re-

searchers and knowledge managers for indexing, re-

trieving and organising data in agricultural informa-

tion systems, and seems to be less used by IT devel-

opers, as discussed by (Baker et al., 2019).

CGIAR - Big Data in Agriculture. is a global

initiative of Semantic Web technology for agricul-

ture, with datasets connecting experts around the

world. The CGIAR (Consultative Group on Inter-

national Agricultural Research) platform companies

apply big data approaches to agricultural use cases.

CGIAR supports selected ontologies across domains,

including the Crop Ontology, the Agronomy Ontol-

ogy (AGRO), the Environment Ontology (ENCO),

the Plant Ontology (PO) and the Socio-economic On-

tology (Arnaud et al., 2020). It stimulates the ex-

change of knowledge between stakeholders. It has

made its 40 separate open data and publication repos-

itories discoverable in one place through GARDIAN,

a data harvesting and annotation service. The GAR-

DIAN ecosystem has expanded to include data from

the US Department of Agriculture, the Indian Coun-

cil of Agricultural Research, the World Bank’s Micro-

data Library and USAID Development Data. Linked

stewardship should enable broader data discovery. 15

CGIAR research centres and 12 research programmes

are partners in the platform, along with 70 external

partners from research to industry, such as Bayer and

Amazon. It covers work from public to industrial,

from analysis to ICT deployment.

FoodOn - for Food Tracebility. Created in 2019,

FoodOn is aimed at ”food traceability”. It inte-

grates existing ontologies and reuses terms from OBO

Foundry, environmental terms from ENVO, agricul-

tural terms from AGRO, etc. Conversely, FoodOn

terms are reused in a growing list of ontologies.

FoodOn extends traceability standards such as the

GS1 standards that enable traceability in the food sup-

ply chain with agronomic and distribution sensor data

(Dooley et al., 2018). FoodOn is part of the open

source OBO Foundry (Open Biological and Biomedi-

cal Ontology Foundry) of interoperable life science-

oriented ontologies. It supports the goals of FAIR

to annotate and share data across research, govern-

ment and commercial sectors. It provides a vocabu-

lary for nutritional analysis, including chemical food

constituents, which are factors in nutrition, health and

plant and animal agriculture research. It aims to build

a comprehensive and easily accessible global farm-

to-fork food ontology that accurately and consistently

describes foods commonly known in cultures around

the world. Although the FoodOn consortium includes

industry partners, it appears to be focused on research

data. Much of FoodOn’s vocabulary comes from

the transformation of LanguaL, a mature and popu-

lar food indexing thesaurus, into an OWL vocabulary

that provides system interoperability, quality control,

and software-driven intelligence.

SAREF4Agri. created in 2019, is the “Smart Agri-

culture and Food Chain” domain extension of SAREF

(Smart Applications and Reference Ontology). This

ontology is maintained by the European Telecom-

munications Standards Institute (ETSI). It speciﬁes

recurring core concepts, main relationships between

concepts and axioms to constrain the use of these con-

cepts and relationships. It is based on a) reuse and

alignment of concepts and relationships deﬁned in ex-

isting assets b) modularity to allow separation and re-

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

351

combination of parts of the ontology depending on

speciﬁc needs and c) extensibility for growth d) main-

tainability to identify and correct defects, accommo-

date new requirements and cope with changes. It has

a SPARQL endpoint and can be downloaded in vari-

ous formats.

GACS - Global Agricultural Concept Scheme.

was created in 2014 as a terminology hub, in-

corporating terms from the three main sources:

AGROVOC from the FAO, CABT (CAB Thesaurus)

from the Centre for Agriculture and Biosciences In-

ternational (CABI) and NAL Thesaurus (NALT) from

the National Agricultural Library of the USA as a

lightweight ”shared concept scheme” (GACS, 2024).

GACS includes in its interoperable concepts identi-

ties related to agriculture from AGROVOC (32,000

concepts), the CAB (140,000 concepts) and the NAL

(53,000 concepts). According to (Baker et al., 2019),

GACS was motivated by the fact that the major the-

sauri were too detailed in terms of custom relation

properties, so that they couldn’t be fully exploited.

They lacked reasoning tools, so it was unclear to de-

velopers what purpose their relationships served. Al-

though they provided standards for data exchange, de-

velopers were unable to extract the knowledge (Baker

et al., 2019). GACS consisted of the most commonly

used (i.e., important) concepts in agriculture (Short

et al., 2023). GACS included an interoperable layer

that transformed data silos into a more reusable web

of data, making resources more discoverable, and its

relative simplicity was intended to make it cheaper

to create and maintain (Baker et al., 2019). How-

ever, major ontologies such as FoodOn, AGRO and

CROP were no longer mapped to GACS, and main-

tenance of GACS was discontinued in 2019. Map-

ping required too many resources, process concepts

were mapped algorithmically, but results were man-

ually checked and inconsistencies were resolved by

discussion. GACS lives on in VocBench for manual

editing and maintenance. In 2022 it was integrated

into NALT as a resource for the “machine age” and is

now a multi-issue concept space with added structural

features for machine readability.

3.2 Ontology Repositories

The number of semantic artefacts is increasing, they

are scattered in different formats with different struc-

tures and from overlapping domains. This is where

AgroPortal comes in as a repository for ontologies

and semantic artefacts (Jonquet et al., 2018). In Jan-

uary 2024, there are 160 ontologies and semantic arte-

facts in AgroPortal (AgroPortal, 2024).

AgroPortal. is conceived as a one-stop shop and

as a “home of ontologies and semantic artefacts in

agri-food and related domains”. It was initiated by

French institutes such as INRAE (French National

Research Institute for Agriculture, Food and Environ-

ment) etc. and involves major international organisa-

tions such as FAO, CGIAR etc. IT collaborates with

Stanford’s BMIR (Center for Biomedical Informatics

Research) in the USA and the EU research infrastruc-

ture consortium LifeWatch within the OntoPortal Al-

liance. (Jonquet et al., 2018; Drury et al., 2019; Jon-

quet et al., 2023). Ontologies and semantic artefacts

can be uploaded. In January 2024, its statistics count

160 ontologies, 59 projects and 409 users, collected

over the last 11 years. AgroPortal provides SPARQL

endpoints as an interface for submitting RDF queries.

AgroPortal provides browsing of features of related

ontologies, it summarises information of ontologies

e.g. calculates a FAIR score. It contains mappings

of classes and concepts. AgroPortal and has contin-

ued the work of GACS. It provides “recommenda-

tions” for relevant ontologies from a text excerpt or

a list of keywords, one can “annotate” ontologies and

sort ontologies by the formats. However, in January

2024, there are only 409 registered users. The most

popular ontologies in AgroPortal in February 2024

are AGROVOC and DEMETER AIM. (DEMETER,

2023) The developers of AgroPortal are actively in-

volved in AGROVOC as users and editors. The in-

ternational ontology community is small and inter-

connected, and also provides OntoPortal as a generic

technology for building ontology repositories. The

AgroPortal researchers at INRAE, namely Clement

Jonquet and others, have provided software for cre-

ating ontology repositories that are domain-agnostic,

customisable and open. The “Ontology Portal Al-

liance” presents (Jonquet et al., 2023) software for

creating ontology repositories and semantic artefact

catalogues that can support resources ranging from

SKOS thesauri to RDF-S and OWL ontologies. The

AgroPortal software is based on OntoPortal, which

emphasises the importance of ontologies and seman-

tic artefacts being FAIR, and seeks to counter a trend

that has been observed, namely the proliferation of

other semantic artefact catalogues, often developed

with their own ad hoc technology and code.

D2KAB-Project (2019-2023). The research team

of LIRMM (Laboratory of Informatics, Robotics and

Microelectronics), INRAE and others responsible for

AgroPortal and OntoPortal are running the D2KAB

project transforming “Data to Knowledge in Agron-

omy and Biodiversity” (D2KAB, 2023) to develop

methods and technologies for ontology lifecycle and

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

352

alignment, to build a “Linked Open Data Cloud”

for agronomy and biodiversity, enabling semantically

driven agronomy and biodiversity science. The focus

is on ontology-based services, ontology alignment as

in GACS, and on building a knowledge graph for agri-

culture and biodiversity. Within D2KAB functionali-

ties have been developed (e.g. SKOS support, FAIR-

ness assessment, interoperability mapping, etc.).

3.3 EU Research on Agricultural

Digital Integration Platforms

To date, ontologies have mainly focused on the clear

deﬁnition of domain concepts and terms in the form

of a vocabulary for annotating research publications

or research datasets. Only a few projects have reused

existing semantic standards within a technological IT

architecture to enable limited data exchange between

different agri-food stakeholders. The use of seman-

tic standards from industry organisations for infor-

mation exchange, such as the GS1 EPICS standard,

ISO (International Organisation for Standardisation),

“AgGateway”, “agrirouter”, does not appear to be

widespread or even beginning to be adopted in re-

search (Brewster et al., 2023). However, the ISO

website states that a new ISO standard for agricul-

ture and food is in preparation. In the development

community, there appears to be little use of ontolo-

gies in the context of sharing and exchanging data

across the agri-food chain, for example for traceabil-

ity and data analysis. In the challenge to build “Agri-

cultural Digital Integration Platforms”, some develop-

ers are now exploring data models derived from on-

tologies as an essential element for semantically in-

teroperable agri-food platforms. Key questions for

research projects in this context are how to improve

the linking of different resources and, secondly, from

a developer perspective, to what extent do the ad-

vantages outweigh the disadvantages of using data

models to share data for service development. In

2019, the EU invested around C80 million in projects

for “Agricultural Digital Integration” (EC, 2020b).

The projects experiment with different approaches

to technical and semantic interoperability, including

the use of data models, service and data manage-

ment approaches involving many partners and pilots.

They applied RAs and incorporated data space con-

cepts. They developed and integrated service com-

ponents of the FMIS, involving many services and

stakeholders in different roles, depending on the ser-

vice, data or process chain orientation. They reused

several FIWARE solutions, e.g. for authentication

or access control. The central infrastructure compo-

nents are currently located (Wolfert et al., 2023) on

the IT ”transformation ladder” between ”data plat-

forms” and ”data spaces”. Projects that exemplify

this challenge are mentioned below, namely CYBELE

(CYBELE, 2022), PLOUTOS (PLOUTOS, 2023),

DEMETER (DEMETER, 2023) and ATLAS (AT-

LAS, 2023), as well as the SmartAgriHubs network

(SmartAgriHubs, 2022) to promote the technologi-

cal approaches described above (EC, 2023c). For an

overview of the state of the art considered in the pro-

posed solution of each EU agritech project, see (Rous-

saki et al., 2023).

ATLAS-Project (2020-2023). is a “service-

oriented solution for interoperability” that solves

the problem without a semantic data model. The

“Agricultural Interoperability and Analysis System”

integrates FMIS with sensor systems and data

analysis on agricultural equipment. The technology

connects ATLAS-enabled systems and establishes

a data ﬂow between them. ATLAS is a distributed,

decentralised network of systems. Each partici-

pating system remains independent, based on its

own technical infrastructure. Semantic interoper-

ability is achieved at the “service level” through

standardised services described in the “ATLAS

Service Templates”. They deﬁne speciﬁcations of

abstract elementary agricultural processes with clear

semantics and deﬁne vendor and technology-agnostic

formal speciﬁcations of the APIs and data formats

to which associated ATLAS services must conform.

Templates allow a service consumer to communicate

with any compliant ATLAS service without requiring

vendor-speciﬁc code, allowing farmers to choose

the compliant ATLAS service that best meets their

speciﬁc needs in terms of quality/accuracy, geo-

graphic/crop speciﬁcity, budget, etc. Formal API

speciﬁcations claim that any client code designed

to integrate the functionality of an ATLAS service

template will interoperate with any ATLAS service

implementing that template. However, this approach

places the burden of the required conversion work

on the side of the customers dealing with the service

templates.

CYBELE-Project (2019-2022). stands for ”Foster-

ing Precision Agriculture and Livestock Farming” is

a semantic metamodel that is conceptualised from use

cases and it is based on general “W3C metadata stan-

dards” to facilitate discovery, exploration, integration

and access to data, with “cross-domain models” such

as the Semantic Sensor Network (SSN) ontology and

“domain-speciﬁc models” such as AGROVOC. CY-

BELE assumes that existing vocabularies can individ-

ually express most of the model concepts, but so far

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

353

there is no single model that satisﬁes all the require-

ments derived from the identiﬁed use cases. In ad-

dition, the merged metamodel deﬁnes concepts that

facilitate access to and querying of data stored in

databases. The model has been tested and claims

to be open to support further structured or semi-

structured data (e.g. JSON) stored in different types

of databases, e.g. noSQL or graph databases. The

use of different types of databases is solved by fol-

lowing linked data approaches to transform or query

different sources on the ﬂy in an integrated manner.

The proposed model has been shown to achieve inter-

operability and homogeneous access to data sources.

However, it could also be used in other domains with

similar data and requirements (Zeginis et al., 2023).

The validation of this approach still needs to be done

(Roussaki et al., 2023).

PLOUTOS-Project (2020-2023). stands for “Data-

driven sustainable agri-food value chains” and uses a

data sharing architecture based on the reuse of ex-

isting semantic standards from different ontologies

and an architecture that computes data along a supply

chain and guarantees digital sovereignty. The project

deﬁnes a “PLOUTOS Core Semantic Model” based

on use cases to achieve data sharing along the sup-

ply chain. The system will allow queries to be made

across a federated network of agri-food stakeholders

(Brewster et al., 2023) that are identiﬁed as common

to the supply chain as a whole. The supply chain

data sharing architecture formalises the GS1 EPICS

standards as ontologies and enables SPARQL queries

across distributed triple stores. The classes and rela-

tions were selected from the requirements of the pi-

lot projects. The data model is based on RDF and

OWL to provide ﬂexibility for modular reuse or ex-

tension. The architecture is based on so-called “Inter-

operability Enablers” (PIEs), which use “graph query

patterns” to traverse the network and collect the re-

quired data to be shared. A core part of the PIE is

the “Knowledge Mapper”, which enables the interop-

erability of the data in the value chain at the seman-

tic level by providing a data translation service that

translates data streams provided by the hosting sys-

tem into selected standardised data formats and vice

versa. The translation functionality of the Knowledge

Mapper is customisable. The data sharing approach

promises to be extensible, but it’s unclear how many

resources this will require. The full version of the

PCSM is available from the GitLab repository.

DEMETER-Project (2020-2023). is building an

“Interoperable, Data-Driven, Innovative, Sustainable

European Agri-Food Sector”. It’s based on a fed-

erated RA referencing IDSA, capable of integrating

the communication, sensing and computing technolo-

gies used in agritech scenarios. It proposes an “Agri-

cultural Information Model” (AIM) as a core service

to enable semantic interoperability between different

FMIS. AIM uses a layered, modular approach that

extends the CYBELE approach. AIM provides pre-

deﬁned mappings between legacy models such as FI-

WARE, Safe4Agri, ADAPT (AgGateway’s ADAPT

Framework), AGROVOC, INSPIRE and Earth Ob-

servation standards. AIM is built on layers (Bochtis

et al., 2022; Roussaki et al., 2023):

• AIM core metamodel following the NGSI-LD

standard (ETSI) as IM approach and API.

• The “cross-domain ontology” layer using a set

of relevant cross-domain standards such as Time

Ontology , SOSA/SSN, GeoSparql (Geographic

Query Language for RDF Data), etc.

• The “domain-speciﬁc ontology” layer modelling

agricultural concepts such as crops, animals,

products, farms etc.

• The “pilot-speciﬁc ontologies layer” for use cases.

• The “metadata schema layer” refers to meta-

information based on cross-domain and domain-

speciﬁc ontologies.

AIM is being tested in 20 pilot projects in 18 coun-

tries. It is said to be easy for developers to use. It is

uploaded to AgroPortal where it is the second most

visited after AGROVOC. It is being standardised by

the OGC (Open Geospatial Consortium). IT develop-

ers are invited to evaluate it. Its successor is DIVINE

(DEVINE, 2023), which demonstrates the costs and

added value of sharing agricultural data.

Research Data Infrastructures (RDIs). provide

resources and services to research communities to

conduct research and foster innovation in their ﬁelds.

They include major equipment or instruments-sets

and knowledge-related facilities such as collections,

archives or scientiﬁc data infrastructures such as on-

tologies or semantic artefacts. As part of the EU

Strategic Plan for Research Infrastructures, RDI will

contribute to achieving the key strategic orientations

of the Horizon Europe Strategic Plan by consolidating

and developing a European Research Infrastructures

landscape and by opening up, integrating and inter-

connecting research infrastructures. As part of the

research infrastructure landscape, the EU maintains

a complete list of Horizon 2020 research infrastruc-

tures by domain, including some food and agriculture

research facilities. In this context, RDI initiatives at

national level are a concrete area of intervention. In

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

354

agriculture in Germany, the NFDI4Agri initiative can

be seen as a ﬁrst step in the ﬁeld of agrosystems re-

search, which operates the FAIRagronet (EC, 2023d).

4 CONCLUSION AND FUTURE

WORK

The paper discusses the current state of IT-based data

sharing solutions in public agricultural research in Eu-

rope, highlighting key sections that play a crucial role

in advancing this ﬁeld. First, the paper presented

the current state of service development for IT-based

data sharing solutions in public agricultural research

in Europe hereby highlighting the need for IT-based

solutions. Based on this need, the following sec-

tions presented solutions for technical interoperability

in the development of federated service and data ar-

chitectures, showing that the agricultural sector pro-

vides a large number of rich and constantly evolv-

ing semantic resources that are organised in a free

and open way. Their development has been mainly

driven by academia and is accompanied by standards

and service solutions for data representation. As a re-

sult, agritech developers can now make general use

of them, for example to reason and deduce on a log-

ical basis. Based on these semantic resources and

common interoperability standards, current European

research projects are developing and evaluating ex-

tensible and layered agricultural information models

as standardised components for Agricultural Digital

Integration Platforms for broad service development.

They reuse and combine standards and best practices

from existing ontologies that are suitable for a wide

range of common use cases and cross-domain require-

ments, promising ease of use and a shortcut to service

development. Nevertheless, the shift of workload to

data providers needs to be carefully considered when

evaluating the beneﬁts of Semantic Web technologies

for service development.

Future research could focus on two key areas: First,

there is a need to further investigate how artiﬁcial

intelligence can improve the use of semantic web

technologies in the agricultural sector. By explor-

ing innovative solutions that automate semantic an-

notation processes using AI algorithms, stakeholders

can streamline data integration, improve knowledge

representation and enhance information retrieval ca-

pabilities within agricultural systems. In that sense,

the use of AI for semantic annotation seems to be

promising, but, further research is needed to be able to

make well-founded decisions. Secondly, it is crucial

to bridge the current research gap between academia

and industry. Investigating ways to facilitate knowl-

edge transfer and collaboration between the agricul-

tural and research sectors can ensure that research

outcomes are effectively translated into practical so-

lutions for the industry and vice versa.

ACKNOWLEDGEMENTS

This work was created as part of the research and

development project ”Stadt-Land-Fluss (SLF)”. The

project with the funding code 2821KI002 is funded by

the German Federal Ministry of Food and Agriculture

on the basis of a resolution of the German Bundestag.

REFERENCES

AgriDataSpace (2022-2024). Building a European frame-

work for the secure and trusted data space for agri-

culture. https://agridataspace-csa.eu/. [Online; last

access May 9, 2024].

AgriGaia (2023). KI-

Okosystem f

ur die mittelstands-

gepr

agte Agrar- und Ern

ahrungsindustrie in

Ankn

upfung an GAIA-X. https://www.agri-gaia.de/

agri-gaia/das-oekosystem/. [Online; last access May

9, 2024].

AgroPortal (2024). AgroPortal. https://agroportal.lirmm.fr.

[Online; last access May 9, 2024].

Arnaud, E. et al. (2020). The Ontologies Community of

Practice: A CGIAR Initiative for Big Data in Agrifood

Systems. Patterns, 1(7):100105.

ATLAS (2019-2023). Agricultural Interoperability and

Analysis System. https://cordis.europa.eu/project/id/

857125. [Online; v May 9, 2024].

Baker, T. et al. (2019). Global agricultural concept space:

lightweight semantics for pragmatic interoperability.

npj Science of Food, 3(1):16. Number: 1 Publisher:

Nature Publishing Group.

Berners-Lee, T. (2006). Design Issues Linked Data. https://

www.w3.org/DesignIssues/LinkedData. [Online; last

access May 9, 2024].

Bochtis, D. D. et al., editors (2022). Information and Com-

munication Technologies for Agriculture—Theme III:

Decision, volume 184 of Springer Optimization and

Its Applications. Springer International Publishing,

Cham.

Brewster, C. et al. (2023). Data sharing in agricultural

supply chains: Using semantics to enable sustainable

food systems. Semantic Web, pages 1–31.

CABI (2024). CABI Improving lives by solving problems

in agriculture and the environment. https://www.cabi.

org/. [Online; last access May 9, 2024].

Caracciolo, C., Stellato, A., Morshed, A., Johannsen, G.,

Rajbhandari, S., Jaques, Y., and Keizer, J. (2013). The

AGROVOC Linked Dataset. Semantic Web, 4(3):341–

348.

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

355

CYBELE (2019-2022). Fostering Precision Agriculture and

Livestock Farming. https://cordis.europa.eu/project/

id/825355. [Online; last access May 9, 2024].

D2KAB (2019-2023). Data to Knowledge in Agronomy

and Biodiversity. https://d2kab.mystrikingly.com/.

[Online; last access May 9, 2024].

DEMETER (2023). Building an Interoperable, Data-

Driven, Innovative, Sustainable European AgriFood

Sector. https://cordis.europa.eu/project/id/857202.

[Online; last access May 9, 2024].

Deutschlandfunk (2024). Großdemo in Berlin. Warum

die Bauern auf der Straße sind. https://www.

deutschlandfunk.de/bauernproteste-102.html. [On-

line; last access May 9, 2024].

DEVINE (2023). Demonstrating value of agridatasharing

for boosting data economy in Agriculture,. https:

//cordis.europa.eu/project/id/101060884. [Online; last

access May 9, 2024].

DFKI (2014). iGreen. http://www.igreen-projekt.de/

iGreen/. [Online; last access May 9, 2024].

Dooley, D. M. et al. (2018). FoodOn: a harmonized food

ontology to increase global food traceability, quality

control and data integration. npj Science of Food,

2(1):23. Number: 1 Publisher: Nature Publishing

Group.

Drury, B., Fernandes, R., Moura, M.-F., and De An-

drade Lopes, A. (2019). A survey of semantic web

technology for agriculture. Information Processing in

Agriculture, 6(4):487–501.

orr, J., Favier, M.-A., and M

ohrle, F. (2023). Inter-

operabilit

at im Digital Farming – Status quo und

zuk

unftige Herausforderungen. Wirtschaftsinformatik

& Management.

EC (2020a). A European strategy for data.

https://eur-lex.europa.eu/legal-content/EN/TXT/

PDF/?uri=CELEX:52020DC0066. [Online; last

access May 9, 2024].

EC (2020b). Agricultural digital integration plat-

forms. https://cordis.europa.eu/programme/id/

H2020 DT-ICT-08-2019/en. [Online; last access

May 9, 2024].

EC (2020c). International Data Spaces Association. https://

internationaldataspaces.org/. [Online; last access May

9, 2024].

EC (2020d). What is Gaia-X. https://gaia-x.eu/

what-is-gaia-x/. [Online; last access May 9, 2024].

EC (2023a). Agri-Digital Conference. https:

//agriculture.ec.europa.eu/eu-agri-food-days/

eu-agri-food-days-conference-material\ en\

#ref-8-december-agri-digital-conference. [On-

line; last access May 9, 2024].

EC (2023b). Data Act Council adopts new

law on fair access to and use of data.

https://www.consilium.europa.eu/en/press/press-

releases/2023/11/27/data-act-council-adopts-new-

law-on-fair-access-to-and-use-of-data/. [Online; last

access May 9, 2024].

EC (2023c). Digitalisation of the European Agri-

cultural Sector: Activities in Horizon 2020.

https://digital-strategy.ec.europa.eu/en/policies/

digitalisation-agriculture-horizon-2020. [Online; last

access May 9, 2024].

EC (2023d). FAIRagro. https://fairagro.net/en. [Online;

last access May 9, 2024].

EC (2024). Interoperable Delivery of European eGov Ser-

vices to Public Administrations Businesses and Citi-

zens. https://ec.europa.eu/idabc. [Online; last access

May 9, 2024].

EC, J. R. C. (2023e). Digital transition: long term implica-

tions for EU farmers and rural communities. Publica-

tions Ofﬁce of the European Union, LU.

EC, DG Informatics (2017). New European interoperabil-

ity framework promoting seamless services and data

ﬂows for EU public administrations. Publications Of-

ﬁce, LU.

EU (2023). Regulation (EC) 2023/2854 of the European

Parliament and of the council of 13.12.2023 on har-

monised rules on fair access to and use of data and

amending Regulation (EU) 2017/2394 and Directive

(EU) 2020/1828 (Data Act). Technical report.

FAO (1981). Multilingual Thesaurus. https://agrovoc.fao.

org/browse/agrovoc/en/. [Online; last access May 9,

2024].

FAO (2021). The State of Food and Agriculture 2021. FAO.

FIspace (2013-2015). Future Internet Business Collabo-

ration Networks in Agri-Food, Transport and Logis-

tics. https://cordis.europa.eu/project/id/604123. [On-

line; last access May 9, 2024].

FIWARE (2018). Smart Agrifood - The new revolu-

tion in farming. https://www.ﬁware.org/community/

smart-agrifood. [Online; last access May 9, 2024].

FIWARE (2024). FIWARE About us. https://www.ﬁware.

org/about-us/. [Online; last access May 9, 2024].

FoodOn (2018-2024). FoodOn: A farm to fork ontology.

https://foodon.org/. [Online; last access May 9, 2024].

GACS (2014-2024). Global Agricultural Concept Scheme.

https://agroportal.lirmm.fr/ontologies/GACS. [On-

line; last access May 9, 2024].

Hoekstra, R. (2009). Ontology representation - design pat-

terns and ontologies that make sense.

Jonquet, C. et al. (2018). AgroPortal: A vocabulary and on-

tology repository for agronomy. Computers and Elec-

tronics in Agriculture, 144:126–143.

Jonquet, C. et al. (2023). Ontology Repositories and Se-

mantic Artefact Catalogues with the OntoPortal Tech-

nology. In The Semantic Web – ISWC 2023, vol-

ume 14266, pages 38–58. Springer Nature Switzer-

land, Cham. Series Title: Lecture Notes in Computer

Science.

LeanIX (2023). Guide Referenzarchitekturen. https://www.

leanix.net/de/wiki/ea/reference-architecture. [Online;

last access May 9, 2024].

NaLamKI (2023). Nachhaltige Landwirtschaft mit

unstlicher Intelligenz. https://nalamki.de/. [Online;

last access May 9, 2024].

NALT (2022-2024). NAL Agricultural Thesaurus. https:

//agclass.nal.usda.gov/. [Online; last access May 9,

2024].

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

356

Ontology, C. (2024a). Crop Ontology for agricultural data.

https://cropontology.org/. [Online; last access May 9,

2024].

Ontology, G. (1978-2024b). Gene Ontology Unifying Bi-

ology. https://www.geneontology.org/. [Online; last

access May 9, 2024].

PLOUTOS (2019-2023). Data-driven sustainable agri-

food value chains. https://cordis.europa.eu/project/id/

101000594. [Online; last access May 9, 2024].

Rodriguez, M. A., Cuenca, L., and Ortiz, A. (2018). FI-

WARE OS Standard Platform in Smart Farming - A

Review. In Camarinha-Matos et al., editors, Col-

laborative Networks of Cognitive Systems, volume

534, pages 581–589. Springer International Publish-

ing, Cham. Series Title: IFIP Advances in Informa-

tion and Communication Technology.

Roussaki, I. et al. (2023). Building an interoperable space

for smart agriculture. Digital Communications and

Networks, 9(1):183–193.

Short, N. M., Woodward-Greene, M. J., Buser, M. D., and

Roberts, D. P. (2023). Scalable Knowledge Manage-

ment to Meet Global 21st Century Challenges in Agri-

culture. Land, 12(3):588. Number: 3 Publisher: Mul-

tidisciplinary Digital Publishing Institute.

SmartAgriFood (2013-2015). Smart Food and Agribusi-

ness: Future Internet for Safe and Healthy Food

from Farm to Fork. https://cordis.europa.eu/project/

id/285326. [Online; last access May 9, 2024].

SmartAgriHubs (2018-2022). Unleashing the innovation

potential for the digital transformation of the Euro-

pean Agrifood Sector. https://www.smartagrihubs.eu/.

[Online; last access May 9, 2024].

Stellato, A., Fiorelli, M., Turbati, A., Lorenzetti, T., van

Gemert, W., Dechandon, D., Laaboudi-Spoiden, C.,

Gerencs

er, A., Waniart, A., Costetchi, E., and Keizer,

J. (2020). VocBench 3: A collaborative Semantic Web

editor for ontologies, thesauri and lexicons. Semantic

Web, 11(5):855–881. Publisher: IOS Press.

Subirats-Coll, I. et al. (2022). AGROVOC: The linked data

concept hub for food and agriculture. Computers and

Electronics in Agriculture, 196:105965.

Subramanian, C., Liu, M., Khan, N., Lenchner, J., Amar-

nath, A., Swaminathan, S., Riegel, R., and Gray,

A. (2024). A Neuro-Symbolic Approach to Multi-

Agent RL for Interpretability and Probabilistic Deci-

sion Making. arXiv:2402.13440 [cs].

Wilkinson, M. D. et al. (2016). The FAIR Guiding Princi-

ples for scientiﬁc data management and stewardship.

Scientiﬁc Data, 3(1):160018.

Wolfert, S. et al. (2021). Navigating the Twilight Zone:

Pathways towards digital transformation of food sys-

tems. Publisher: Wageningen University & Research.

Wolfert, S. et al. (2023). Digital innovation ecosystems in

agri-food: design principles and organizational frame-

work. Agricultural Systems, 204:103558.

Zeginis, D. et al. (2023). A semantic meta-model for data

integration and exploitation in precision agriculture

and livestock farming. Semantic Web, pages 1–29.

Use of Semantic Artefacts in Agricultural Data-Driven Service Development

357