=Paper=
{{Paper
|id=Vol-3184/MK_short3
|storemode=property
|title=Ontological Representation of Cultivated Plants: Linking Botanical and Agricultural Usages
|pdfUrl=https://ceur-ws.org/Vol-3184/MK_short3.pdf
|volume=Vol-3184
|authors=Baptiste Darnala,Florence Amardeilh,Catherine Roussey,Konstantin Todorov,Clément Jonquet
|dblpUrl=https://dblp.org/rec/conf/esws/DarnalaARTJ22
}}
==Ontological Representation of Cultivated Plants: Linking Botanical and Agricultural Usages==
https://ceur-ws.org/Vol-3184/MK_short3.pdf
Ontological Representation of Cultivated Plants:
Linking Botanical and Agricultural Usages⋆
Baptiste Darnala1,2 , Florence Amardeilh2 , Catherine Roussey3 , Konstantin Todorov1
and Clément Jonquet1,4
1
LIRMM, University of Montpellier, CNRS, Montpellier, France
2
Elzeard, Cité du Numérique, Bègles, France
3
Université Clermont Auvergne, INRAE, UR TSCF, F-63000 Clermont–Ferrand, France.
4
MISTEA, University of Montpellier, INRAE, Institut Agro, Montpellier, France
Abstract
Cultivated plants may be described from various viewpoints: botanical, agronomic, agricultural and
more. These viewpoints often result into different specific formal representations (i.e., ontologies).
Linking concepts describing these different viewpoints is difficult and demands domain expertise. Still
it is necessary as it supports the agricultural planning processes. In our case, there exists no standard
knowledge pattern to represent alignments between thesauri describing cultivated plants in agriculture
and organism taxonomies or classifications; in addition, basic ontology mapping properties (e.g., from
SKOS) are not sufficient. We have conceived the Crop Planning and Production Process Ontology
(C3PO) to describe agricultural knowledge for diversified crop production. In this paper, we describe the
ontological representation of the Plant module of C3PO, which addresses the aforementioned linking
needs. It integrates crop usage information about cultivated plants from the French Crop Usage thesaurus
and botanical –classification and nomenclature– information from the TaxRef taxonomy. This Plant
module is valued in two systems both developed by Elzeard—a french SME: (i) a web application to
support farmers in crop planting activity; and (ii) a web portal, La Serre des Savoirs (under development),
which will share general agricultural information about crops.. The C3PO ontology and its knowledge
graph are publicly available at https://gitlab.com/serre-des-savoirs/c3po.
Keywords
knowledge graphs, ontologies, ontology modelet, knowledge integration, agriculture, botanic taxonomy
1. Introduction
A plant is a complex system studied and observed by different experts (biologists, agronomists,
botanists, farmers) who each use specific characteristics to describe it. Each of these viewpoints
are captured into several ontologies or knowledge graphs (KG) such as: (i) TAXREF-LD [7],
which represents –as linked data– the national repository on fauna and flora of metropolitan
INTERNATIONAL WORKSHOP ON KNOWLEDGE GRAPH GENERATION FROM TEXT (TEXT2KG 2022) and MODULAR
KNOWLEDGE (2022), ESWC 2022, Hersonissos, Greece, May 29, 2022
Envelope-Open baptiste.darnala@elzeard.co (B. Darnala); florence.amardeilh@elzeard.co (F. Amardeilh);
catherine.roussey@inrae.fr (C. Roussey); konstantin.todorov@lirmm.fr (K. Todorov); jonquet@lirmm.fr
(C. Jonquet)
Orcid 0000-0001-7390-4850 (B. Darnala); 0000−0002−6306−4437 (F. Amardeilh); 0000−0002−3076−5499 (C. Roussey);
0000-0002-9116-6692 (K. Todorov); 0000−0002−2404−1582 (C. Jonquet)
© 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
CEUR Workshop Proceedings (CEUR-WS.org)
France and overseas territories; (ii) the French Crop Usage (FCU) [11] which represents a
thesaurus of cultivated plant organised by agriculture usages in France (human food, industry,
cattle feed).1 Neither TAXRED-LD, nor FCU include all the plant characteristics into a unique
knowledge representation. No semantic resource actually provide a fully integrated and unified
view, making alignment/linking between such resources mandatory. Indeed, all this information
is important for agriculture when farmers struggle to plan and optimise crop production.
Elzeard (https://elzeard.co)—a French SME which develops an application dedicated to farmers
(market gardeners and vegetable growers)—has conceived a modular ontology called Crop
Planning and Production Process Ontology (C3PO) [3]. C3PO represents plot management
and crop itineraries from an agricultural perspective. It is the backbone of a web application
currently under development called Elzeard application to assist farmers in their planning and
production activities and La Serre des Savoirs 2 a web portal under development to publish the
culture and the crop itinerary publicly as a wiki portal. The domain of activities being extremely
complex, the application requires clear and unambiguous knowledge to support farmers in
their planning choices. C3PO’s sub-part dedicated to cultivated plant representation, the Plant
module describes all the knowledge specific to plant, Elzeard has aggregated and linked to
agricultural production concepts. The module hierarchizes cultivated plants from a farmer
perspective and collects information to describe objects at each different hierarchy level, e.g.
plant/crop/family.
Our need is to get and aggregate data—structured as knowledge graphs—from different
domains: farmers, agronomic and botanical knowledge. In this paper, we focus on the links
between C3PO and TAXREF-LD for the integration of French botanic knowledge, and C3PO
and FCU for the integration of the French agricultural usage knowledge. The difficulty is thus
to align classes/concepts and individuals from different knowledge graphs, where each graph
describes a different viewpoint. For example, farmers use the term ‘solanaceous fruits’ which
somehow was borrowed from the scientific name ‘solanaceae’. The related scientific taxon
groups the plants tomatoes, potatoes, eggplants and peppers. However, from the farmer point
of view, solanaceous never include potatoes because potato farming practices are very different
from the three other ones. Our goal is, therefore, to borrow in La Serre des Savoirs’s knowledge
graph, from scientific taxonomies or referential agricultural usages. Plant groups and families
still coping with the concrete differences observed within the fields. We choose to use double
typing to construct this alternative hierarchies, to link individuals to external knowledge graphs
and to add the specific properties to each instance. As explained later, the elements of the Plant
module’s hierarchy are both instances of o w l : C l a s s and s k o s : C o n c e p t thus can be described
with SKOS properties [8] and inherit properties from OWL descriptions. Plus, it permits to link
with other knowledge graphs either relying on SKOS or OWL.
In this paper, we present C3PO’s Plant module, how it links data from different semantic
resources and how it builds a knowledge graph useful for a specific task, such as agricultural
production. Section 2 covers related work on plant description models. Section 3 explains the
ontological representation of the Plant module of C3PO and its links with TAXREF-LD and
FCU. Section 4 describes the data model and instantiation.
1
https://doi.org/10.15454/QHFTMX
2
‘Greenhouse of knowledge’, in French.
2. Related Semantic Resources for Plants
C3PO is built with Semantic Web technologies, i.e. it provides the integration of ontologies
to model the plants or extract information from open knowledge graphs with facility. Plant
module represents vegetable crops and links them to already existing knowledge graphs and
ontologies. When building C3PO’s Plant module, we searched the term ‘plant’ in the AgroPortal
ontology repository [6] which returned results in multiple ontologies containing plant related
knowledge:
• In the first category, we found resources describing plants within a taxonomy of organ-
isms / biological entities: TAXREF-LD, previously cited and the NCBI Taxonomy which
describes the standard nomenclature and classification of international organism.3
• In the second category, we found resources describing plants in farming usage: FCU
previously cited and the GECO ontology [12], the backbone of an agro-ecological knowl-
edge base to describe new agricultural practices. Plants within GECO being based on a
previous version of FCU.
• In the third category, we found resources presenting an experimental or productive view-
point: FoodOn [4], an exhaustive‘farm-to-fork’ ontology about food related knowledge
which contains several crop descriptions and some specific cultivars; The Agronomy
Ontology [1] an ontology for “representing agronomic practices, techniques, variables
and related entities” which contains a representation of ’crop’; FOODIE [9] an ontology
which represents a monitoring process of one crop on one area at one moment. The
plant is represented by crop species classes. Neither FoodOn, Agronomy Ontology and
FOODIE propose a hierarchy of crops.
• In the last category, we found resources describing composition of plants. Plant Ontology
(PO) [5] describes plant’s characteristics (anatomy, morphology, growth), it is composed
of a structured collection of terms that describe structure and developmental stages of a
plant. Plant Trait Ontology (TO) [2] is a vocabulary that describes phenotypic traits in
plant. These ontologies are used to described crops by the view of an existing unique
plant in the Crop Ontology, a project that describes each crop by a specific ontology.
In these resources, the plant hierarchy is absent or not sufficient for our farming use-case.
The requirement is a hierarchy seen by a farmer that integrates a maximum of information that
can help him in his production. As plants are a mix of all the information describes above, we
need to be able to characterise plants either from a botanical or agronomic point-of-view. To do
this, we need a generic model describing what a farmer understands about plants, which links
and integrates external resources to obtain the most complete depiction.
3. C3PO’s Plant Module Description
C3PO [3] is an ontology composed of several ontology modules, each describing a specific part of
a farm and its processes. The design of plots, the administrative organisation, the management
of cultivation processes, the description of supplies (input, equipment and plant materials), a
3
https://www.ncbi.nlm.nih.gov/taxonomy
Figure 1: Overview of the Crop Planning and Production Process Ontology
sale manager module and one to describe the plants are the main modules. Figure 1 shows an
overview of these ontology modules and their relations. The decomposition of C3PO in several
ontology modules allows us to better manage the complexity of the farming domain.
The plant module describes the plant knowledge for C3PO and creates links to external
resources to improve the plant representation. We proceed to present the specification of the
module and the linking method that we have adopted.
3.1. Specification
In agriculture, according to farmers, a cultivated plant is generally part of a collection, i.e., a
cultivated family. Both are defined as:
Cultivated Plant, a type of plant; the type gathers information about how the farmer will
cultivate all the plants of this type. Examples of plant types are Carrot or Onion.
Cultivated Family, a set of plant types that are grouped based on some plant type character-
istics. The characteristic could be botanical characteristic like the species (Daucus carota),
or usage characteristic like Leaf vegetable.
3.2. Plant linking
We used the SAMOD [10] ontology development agile methodology. We have described the
aforementioned linking knowledge pattern as a modelet i.e. a “A modelet is a stand-alone model
describing a particular domain”. We have focused on linking TAXREF-LD and FCU through
C3PO.
Our requirement is a Plant module which represents a plant hierarchy described by farmers
and merges information from different resources. We therefore created our instances as pivot
objects reifying the links between the resources to merge, these objects being themselves char-
acterized by properties and classes in their own hierarchy. To do this, we choose to represent
C3PO plant instances both as s k o s : C o n c e p t and o w l : C l a s s . SKOS allows the description of the
hierarchy with broader / narrower relations plus linking to external objects with mapping
properties s k o s : * M a t c h . OWL allows to describe knowledge about things and relations between
them. The double typing makes it possible to recover the competences of each one. It also
makes the class generic enough on the level of the modelling to be able to connect informa-
tion coming from external resources which are made in SKOS as in OWL. The hierarchy of
skos:broader/narrower enables to retrieve any element and provides some information retrieval
service.
This modelet is composed of several hierarchized classes specified in Subsection 3.1:
c 3 p o : B o t a n i c a l F a m i l y , c 3 p o : U s a g e F a m i l y and c 3 p o : C u l t i v a t e d P l a n t ; their instances will be
double typed as s k o s : C o n c e p t and will be linked to others resources. CultivatedPlant instances
are linked with BotanicalFamily and UsageFamily instances by a SKOS broader/narrower rela-
tion as Plant are part of a family, like Onion is part of Alliaceae family. SKOS is used to make
the plant organisation as owl:subClassOf is not made to declare a family and its members.
FCU is formalized in SKOS with objects instances of f c u : C r o p , a specialization of s k o s : C o n c e p t .
Therefore, we can link C3PO plant instances and FCU instances with SKOS properties.4 The link
between C3PO and TAXREF-LD is made with a C3PO property called c 3 p o : h a s S c i e n t i f i c N a m e .
C3PO classes are a pivot between the external resources. Figure 2 shows the modelet.
4. Data model and instantiation
C3PO’s knowledge graph contains 118 instances of c 3 p o p l a n t : C u l t i v a t e d P l a n t , current crops
of interest for La Serre des Savoirs’s clients. We were then able to link them manually to FCU
and TAXREF-LD and involve three agronomy experts to validate them. 74 cultivated plants
have both a link to TAXREF-LD and FCU, 6 only to TAXREF-LD, and 36 only to FCU. Figure 3
presents and example of the links for C3PO’s instance Onion_i.
The individual c 3 p o k b : O n i o n - i borrows its scientific name ‘Allium cepa’
from t x r f : n a m e / 8 1 3 3 9 . Those two individuals are linked by the property
c 3 p o p l a n t : h a s S c i e n t i f i c N a m e . The individual c 3 p o k b : O n i o n - i borrows its preferred common
name “Oignon”@fr from t x r f : t a x o n / 8 1 3 3 9 / 1 0 . 0 . Those two individuals are linked by the
property chain composed of c 3 p o p l a n t : h a s S c i e n t i f i c N a m e and t x r f p : h a s R e f e r e n c e N a m e . The
individual c 3 p o k b : O n i o n - i borrows its french preferred common name “Onion”@en from
f c u : O i g n o n s . Those two individuals are linked by the property f c u : h a s R e l a t e d C r o p and
s k o s : e x a c t M a t c h . f c u : h a s R e l a t e d C r o p is declared to link with a f c u : C r o p , declare a s k o s : * M a t c h
add more semantic to the link.
At the end, this choice of modelling extends the C3PO’s knowledge graph and adds more
labels. Moreover, it improves the description by grouping cultivated plants under multiple
representations. For example, the onion of C3POo is under family of Alliacees and the link with
TAXREF-LD extends the family description because the taxon is under Amaryllidaceae family.
These two descriptions come from different botanical representations. As an agricultural advice
4
Later, we will link C3PO instances with other resources such as the NCBI Taxonomy.
Figure 2: Knowledge pattern (modelet) interlinking C3PO, TAXREF-LD and FCU.
Figure 3: Example of instantiation with onion as cultivated plant.
can be based on different representations, having larger knowledge will help farmers in their
research. For example, we consider the following use-cases:
• Inputs management. These are products bought or made by farmers and used for agri-
cultural production, like seeds or fertilizers. For chemical products, there are official
regulations, which may depend on the plant but also on the botanical and usage families.
• Manage the crop rotations. Rotation rules are plant or family dependent. It is important
to know the membership of Cultivated plant in Cultivated family to infer crop succession.
A rotation system will combine this membership and the historical information stored in
the crop and plot management modules.
.
5. Conclusions and Future Work
Linking and integrating data from different resources is a complex process that needs curation.
This paper proposes modelling multiple viewpoints inside a single hierarchy using double typing
with s k o s : C o n c e p t and o w l : C l a s s . This allows to model the hierarchy, and in the meantime
align our instances with external resources and inherit class properties. C3PO Plant module’s
instances are generic enough to be linked to other knowledge graphs such as NCBI or EPPO
Global Database.5
More generally, other scientific domains (legal, health) could display the problem of connect-
ing data organised by different viewpoints and our methodology using a pivot class and double
typing could be considered.
Acknowledgements
We acknowledge support from the National Office for Biodiversity with
MesclunDurab grant and the Nouvelle-Aquitaine Region with ”Social Innovation AMI” and
”Digital Prototypes” grants. This work was also partially achieved with support of the Data to
Knowledge in Agronomy and Biodiversity (D2KAB – www.d2kab.org) project that received fund-
ing from the French National Research Agency (ANR-18-CE23-0017) and the project ”Partages
de Connaissances” (PACON) of the transverse programme MetaBio funded by INRAE. We also
thank Dr. Kevin Morel (INRAE), Matthieu Hirshy (ACTA) and Juliette Raphel (Elzeard) for
curating the alignments as well as all contributors from MesclunDurab and D2KAB projects for
their constructive feedback.
References
[1] Aubert C., Buttigieg P.L., Laporte M.A., Devare M., Arnaud E., (2017) CGIAR Agronomy
Ontology, http://purl.obolibrary.org/obo/agro.owl, licensed under CC BY 4.0.
[2] Elizabeth Arnaud et al. “Towards a reference plant trait ontology for modeling knowledge
of plant traits and phenotypes”. In: International Conference on Knowledge Engineering
and Ontology Development. Vol. 2. SciTePress. 2012, pp. 220–225.
[3] Baptiste Darnala et al. “Crop Planning and Production Process Ontology (C3PO), a
new model to assist diversified crop production”. In: Integrated Food Ontology Workshop
(IFOW’21) at the 12th International Conference on Biomedical Ontologies (ICBO). 2021.
[4] Damion M. Dooley et al. “FoodOn: a harmonized food ontology to increase global food
traceability, quality control and data integration”. In: npj Science of Food 2.1 (Dec. 2018),
p. 23. issn: 2396-8370. doi: 10.1038/s41538-018-0032-6.
[5] Pankaj Jaiswal et al. “Plant Ontology (PO): a controlled vocabulary of plant structures
and growth stages”. In: Comparative and functional genomics 6.7-8 (2005), pp. 388–397.
5
https://gd.eppo.int
[6] C. Jonquet et al. “AgroPortal: an ontology repository for agronomy”. In: Computers and
Electronics in Agriculture 144 (2018), pp. 126–143.
[7] F. Michel et al. “A Model to Represent Nomenclatural and Taxonomic Information as
Linked Data. Application to the French Taxonomic Register, TAXREF”. In: Proceedings
of ISWC 2017 Workshop on Semantics for Biodiversity (S4Biodiv 2017), Oct 2017, Vienna,
Austria (2017), pp. 1–12.
[8] Alistair Miles and Sean Bechhofer. “SKOS simple knowledge organization system refer-
ence”. In: (2009).
[9] Raúl Palma et al. “An INSPIRE-based vocabulary for the publication of Agricultural
Linked Data”. In: International Experiences and Directions Workshop on OWL. Springer.
2015, pp. 124–133.
[10] Silvio Peroni. “A simplified agile methodology for ontology development”. In: OWL:
Experiences and Directions–Reasoner Evaluation. Springer, 2016, pp. 55–69.
[11] Catherine Roussey et al. “A methodology for the publication of agricultural alert bulletins
as LOD”. In: Computers and Electronics in Agriculture 142 (2017), pp. 632–650. issn: 0168-
1699. doi: https://doi.org/10.1016/j.compag.2017.10.022.
[12] V. Soulignac et al. “GECO, the French Web-based application for knowledge management
in agroecology”. In: Computers and Electronics in Agriculture 162 (2019), pp. 1050–1056.