SAREF4AGRI ontology and semantics

Introduction and overview

The present document is a technical specification of SAREF4AGRI, an OWL-DL ontology that extends SAREF [1] for the agricultural domain.

The present document is a major revision of the SAREF4AGRI ontology extension, developed in the context of the STF 641, using updated reference ontology patterns specified in ETSI TS 103 548 [2] to solve the harmonization needs identified in ETSI TR 103 781 [i.5], with updated development framework and tools defined in ETSI TS 103 673 [3].The present document was originally developed in the context of the STF 534, an ETSI specialists task force that was established with the goal to extend SAREF [1] for the domains of Smart Cities, Smart Industry & Manufacturing, and Smart AgriFood (https://portal.etsi.org/STF/stfs/STFHomePages/STF534). The intention of SAREF4AGRI is to connect SAREF with existing ontologies (such as W3C SSN, W3C SOSA, GeoSPARQL, etc.) and important standardization initiatives in the Smart Agriculture and Food Chain domain, including ICAR for livestock data (https://www.icar.org/), AEF for agricultural equipment (http://www.aef-online.org), Plant Ontology Consortium for plants (http://archive.plantontology.org), AgGateway for IT support for arable farming (http://www.aggateway.org/), as mentioned in the associated SAREF4AGRI requirements document ETSI TR 103 511 [i.2].

To show the potential of SAREF4AGRI, the present document focuses on two examples, which are the "livestock farming" and "smart irrigation" use cases. Various other examples exist in the Smart Agriculture and Food Chain domain, such as arable farming, horticulture, agricultural equipment, greenhouses and food chain, as mentioned in [i.2] (for an exhaustive list of use cases, see also the H2020 Large Scale Pilot "Internet of Food and Farm 2020 (IoF2020)" at https://iof2020.eu/trials). However, it was necessary to make actionable choices within the STF 534 timeframe and the available resources, thus livestock farming and smart irrigation have been chosen as the two initial examples to create SAREF4AGRI. As a next step, it is recommended to further refine the proposed livestock farming and smart irrigation examples to add relevant sensors that are not considered yet, and also consider additional use cases to create new releases of SAREF4AGRI, following and extending the examples provided in the present document. As all the SAREF ontologies, SAREF4AGRI is a dynamic semantic model that is meant to evolve over time. Therefore, the stakeholders in the AgriFood domain (starting from the ICAR, AEF and AgGateway initiatives) are invited to use, validate and provide feedback on SAREF4AGRI, collaborating with the SAREF ontology experts to improve and evolve SAREF4AGRI in an iterative and interactive manner, so that changes and additions can be incorporated in future releases of the present document.

The livestock farming and smart irrigation use cases used as basis to create SAREF4AGRI in the present document are concerned with the integration of multiple data sources for the purpose of providing decision support services located on the local "Farm Management System" of the farmers or provided by a service over the network. Multiple data sources of interest include GPS, meteorological data (both historic and current), remote observation (via satellite sources such as Copernicus) and local observation using near or proximal sensors. As an extension of SAREF, which is a semantic model for IoT that describes smart devices and applications in terms of their functions, services, states and measurements [1], SAREF4AGRI is concerned with the description of proximal sensors that measure a variety of relevant parameters for agriculture, including: (on animal) movement, temperature, etc., (in the soil) moisture/humidity, Ph value, salinity, compaction, (on plant) plant colour (NDVI), etc. The measurements from these sensors need to be integrated by a decision support service to enable the planning of (for example) a treatment plan for animals (in a livestock scenario), or a decision to irrigate or harvest (in an irrigation, horticulture or greenhouse context). The requirements used to create the SAREF4AGRI extension specified in the present document are described in the associated ETSI TR 103 511 [i.2].

The prefixes and namespaces used in SAREF4AGRI and in the present document are listed in Table 1.

Diagrams are to be interpreted using the Chowlk notation [3], Clause 9.7.2.

SAREF4AGRI

Void

Platform, System and Deployment

The main entities in the modelling of platforms, systems and deployments are represented by the ssn:System and ssn:Deployment classes. Note that the design patterns for modelling these concepts have been taken from the W3C SSN ontology and, as a best practice for reuse, the SAREF4AGRI model refers directly to the URIs of the SSN (http://www.w3.org/ns/ssn/) and SOSA (http://www.w3.org/ns/sosa/) ontologies.

The ssn:System class in the SSN ontology represents a system and is components as specific devices, actuators or sensors. Moreover, the ssn:Deployment class from the SSN ontology describes the deployment of one or more systems on a sosa:Platform for a particular purpose for a given time period. SAREF4AGRI defines a saref:Device as subclass of an ssn:System and extends the ssn:Deployment class by means of the s4agri:Deployment class. In this way, it is possible to represent a specific installation of a certain agricultural system (e.g. a smart irrigation system) in a given space (expressed by means of the property s4agri:isDeployedAtSpace) and at a given temporal frame (expressed by means of the property s4agri:hasDeploymentPeriod) where SAREF4AGRI devices (e.g. a pluviometer, a soil tensiometer, a weather station, and a watering gun) can be deployed. The deployment can involve a given sosa:Platform which hosts the system deployed in such deployment. In order to represent temporal information the TIME ontology has been reused. For the geographical information the GeoSPARQL ontology (http://www.opengis.net/ont/geosparql#) is reused.

Table 2 summarizes the properties that characterize the s4agri:Deployment class.

Void

Animal, Crop and Soil (Feature of Interest)

The main features of interest in SAREF4AGRI currently support (aspects of) the livestock farming and smart irrigation use cases and are represented by the s4agri:Animal, s4agri:AnimalGroup, s4agri:Crop and s4agri:Soil classes that are shown in Figure 1.

The s4agri:Animal class describes an animal that can be classified in SAREF4AGRI reusing the TAXRANK taxonomy vocabulary (http://purl.obolibrary.org/obo/taxrank.owl#). Besides the reuse of the TAXRANK taxonomy vocabulary, an animal is furthermore defined in SAREF4AGRI as having a birth and death date. An animal also has a unique identifier and can be part of one or more s4agri:AnimalGroup that are used to conduct experiments and observations on the livestock. Note that animals can be also specialized using subclasses, as is shown in the example in clause 4.3.1 with the ex:LactatingCow class that was created as a subclass of s4agri:Animal. Animals and animal groups are related to measurements via the saref:FeatureOfinterest concept of SAREF [1].

The s4agri:Soil class represents the upper layer of the earth in which plants grow. The s4agri:Crop class describes a collection of homogeneous plant species that is grown on a large scale commercially (especially a cereal, fruit, or vegetable) and is planted on a single location. A s4agri:Crop is grown on some s4agri:Parcel, which is an area of land, defined in SAREF4AGRI as subclass of the geo:Feature (see clause 4.2.6). Moreover, s4agri:Crop is related to measurements via saref:FeatureOfInterest [1] .

Table 3 and Table 4 summarize the definitions of the main classes and properties described above.

Device

SAREF4AGRI extends the device hierarchy defined in SAREF in order to include devices needed to support the livestock farming and the smart irrigation use cases. These devices are shown in Figure 2. The devices included for the Smart Irrigation use case are: s4agri:Pluviometer, s4agri:SoilTensiometer, s4agri:WeatherStation, and s4agri:WateringGun. The devices included for the Livestock Farming use case are: s4agri:MovementActivitySensor, EatingActivitySensor, s4agri:MilkingSensor, and s4agri:WeightSensor.

Property

SAREF4AGRI extends the property hierarchy defined in SAREF in order to include properties needed to support the livestock farming and the smart irrigation use cases. These devices are shown in Figure 3.

The properties included for the smart irrigation use case are: s4agri:SoilMoisture, s4agri:IrrigationWater, s4agri:SoilTemperature, s4agri:AirTemperature, s4agri:AmbientHumidity, s4agri:Precipitation, and s4agri:PlantGrowthStage.

The properties included for the livestock farming use case are: s4agri:Yield (which can further be specialized in subclasses, such as MilkYield, CropYield, MeatYield, MilkYield, etc. as needed) and s4agri:Intake (which can further be specialized in subclasses, such as FoodIntake for animals, FertilizerIntake for crops, etc. as needed).

Table 5 summarizes the definitions of the classes described above.

Topology

SAREF4AGRI adopts the same topology modelling pattern that is adopted in the SAREF4CITY extension [i.3], where existing standard ontologies have been reused for this purpose. As shown in Figure 4, for representing spatial objects in SAREF4AGRI, the geo:SpatialObject class from GeoSPARQL has been reused along with its subclasses geo:Feature, geo:Geometry and the properties geo:sfContains, geo:sfWithin and geo:hasGeometry.

For the purpose of SAREF4AGRI, the geo:Feature class has been extended with the following subclasses:

• the s4agri:Farm
• the s4bldg:Building
• the s4bldg:BuildingSpace
• the s4agri:Parcel

A s4agri:Farm can contain one or more s4bldg:Building and s4agri:Parcel (via the geo:sfContains relation). Note that these types of feature are used in the present document as examples, but more feature types (and building types) can be added as needed. Moreover, a s4bldg:Building can be further decomposed in one or more s4bldg:BuildingSpace individuals (once again via the geo:sfContains relation). As subclasses of geo:Feature, all the classes mentioned above inherit the possibility to have a physical geometric description using geo:Geometry, if needed (e.g. especially relevant for s4agri:Parcel).

Person and Organization

The SAREF4AGRI extension reuses the FOAF vocabulary (http://xmlns.com/foaf/0.1/) to represent the concepts of Person and Organization. Figure 5 shows that in SAREF4AGRI the foaf:Person and foaf:Organization classes are extended with the s4agri:Farmer and s4agri:FarmHolding subclasses to describe farmers and their organizations, respectively. Both foaf:Person and foaf:Organization are subclass of foaf:Agent. Organizations (e.g. s4agri:FarmHolding)have members (e.g. farmers). Both s4agri:Farmer and s4agri:FarmHolding can manage some s4agri:Farm.

Instantiating SAREF4AGRI

Livestock farming

This clause shows an example of how to instantiate the SAREF4AGRI extension of SAREF for the livestock farming use case. The example describes a family company owned farm that grows certain crops and owns lactating cows. Various sensors are used in the farm to monitor animals and crops.

The first part of the example is related to the organizational aspects of the farm and is shown in Figure 6.

Figure 6 shows two instances of a farmer, namely ex:H._Jansen and ex:J._Jansen, which are both members of the s4agri:FarmHolding ex:Jansen_and_Sons. The organization manages the s4agri:Farm ex:Farm_Jansen_and_Son_Eindhoven.

Note that the s4agri:Farm is a subclass of geo:Feature, which enables the description of the exact geometrical aspects of the area. Moreover, ex:Farm_Jansen_and_Son_Eindhovencontains the following additional instances of type geo:Feature:

• four s4agri:Parcel individuals: ex:Parcel_South, ex:Parcel_West, ex:Parcel_North, and ex:Parcel_East
• two s4bldg:Building individuals: ex:Milk_Cow_Barn, ex:Heated_Glass_Greenhouse

Furthermore, the figure shows that ex:Parcel_East and ex:Parcel_West both contain some s4agri:Crop (ex:Sweet_Corn_1 and ex:Sweet_Corn_2, respectively). Additionally, ex:Parcel_North contains the s4agri:AnimalGroup individual ex:Cow_Group_A, which consists of the s4agri:Animal individuals ex:Cow1 and ex:Cow2. Finally, Figure 6 shows that ex:Parcel_South does not contain anything.

Figure 7 elaborates on ex:Parcel_North that contains the ex:Cow_Group_A with two cows (i.e. ex:Cow1 and ex:Cow2) which are similarly taxonomically described using the TAXRANK taxonomy vocabulary. The ex:Cow_Group_A generates s4agri:MilkYield, which is a type of s4agri:Yield and consequently a saref:Property. The example contains one instance of s4agri:Milk_Yield that represents the outcome of the milking procedure of a certain cow. The s4agri:Milk_Yield instance is measured in om:Liter by the ex:MilkYieldSensor. The ex:MilkYieldSensor is of type s4agri:MilkingSensor, which is a saref:Sensor, and thus a saref:Device. Figure 7 further shows that the sensor is contained in an ex:Milking_Machine, which is a saref:Device, and the ex:Milking_Machine has a sensor that measures the yield. The measurements are directly linked to the sensor, instead of to the milk machine itself, because a large machine can have multiple sensors.

Figure 8 further shows an example of another s4agri:AnimalGroup, namely ex:Cow_Group_B. This s4agri:AnimalGroup only contains a single cow, ex:Cow3, whose eating activity is being monitored by ex:Cow_Eating_Activity_Sensor_33. This s4agri:EatingActivitySensor made two measurements about the cow eating activity.

Smart Irrigation

This clause shows an example of how to instantiate the SAREF4AGRI extension of SAREF to represent the deployment of some sensors and an example of measurement for the smart irrigation use case. This example is shown in Figure 9.

The ex:ArvalisDeployment20162017Land07 deployment is deployed in the ex:PlatformArvalisLand07 platform and has a deployed system, namely the smart irrigation station ex:ArvalisIrrinovStation01. The deployment takes place in the time interval between January 2016 and the end of 2017, defined as ex:TimeInterval2016-2017.

The deployed system, ex:ArvalisIrrinovStation01, is composed of two sensors identified by the URIs ex:ArvalisIrrinovStation01SoilSensor01 and ex:ArvalisIrrinovStation01SoilSensor02 which are linked from the system by the property ssn:hasSubSystem. Both sensors are of the type ex:SoilTensiometer. Both sensors measure (saref:observes) the soil moisture property (s4agri:SoilMoisture).

A measurement taken by the ex:ArvalisIrrinovStation01SoilSensor02 sensor is also depicted, namely the measurement ex:ArvalisIrrinovStation01SoilSensor02ObservationAtPT24H2016-06-14T000000_0200. This measurement is about (saref:isValueOfProperty) the soil moisture property (s4agri:SoilMoisture). This measurement has a value of 1 490,0 expressed in millibars (om:millibar).

References

Normative references

• [0] ETSI TS 103 410-6 (V2.1.1): "SmartM2M;; Extension to SAREF; Part 6: Smart Agriculture and Food Chain Domain".
• [1] ETSI TS 103 264: "SmartM2M; Smart Applications; Reference Ontology and oneM2M Mapping".
• [2] ETSI TS 103 548: "SmartM2M; SAREF reference ontology patterns".
• [3] ETSI TS 103 673: "SmartM2M; SAREF Development Framework and Workflow, Streamlining the Development of SAREF and its Extensions".

Informative references

• [i.1] ETSI TR 103 411: "SmartM2M Smart Appliances SAREF extension investigation".
• [i.2] ETSI TR 103 511: "SmartM2M; SAREF extension investigation; Requirements for AgriFood domain".
• [i.3] ETSI TS 103 410-4: "SmartM2M; Extension to SAREF; Part 4: Smart Cities Domain".
• [i.4] Verhoosel J. and Spek J.: "Applying Ontologies in the Dairy Farming Domain for Big Data Analysis". Proceedings of the 1^st Semantic Web Technologies for the Internet of Things (SWIT) 2016 workshop, co-located with 15^th International Semantic Web Conference (ISWC 2016), Kobe, Japan, October 2016, pg. 91-100, CEUR.
• [i.5] ETSI TR 103 781: "SmartM2M; Study for SAREF ontology patterns and usage guidelines".

Acknowledgements

The editors would like to thank the ETSI SmartM2M technical committee for providing guidance and expertise.

Also, many thanks to the ETSI staff and all other current and former active Participants of the ETSI SmartM2M group for their support, technical input and suggestions that led to improvements to this ontology.

Also, special thanks goes to the ETSI SmartM2M Technical Officer Guillemin Patrick for his help.