Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major...

19
Interlinked data and models using a semantic approach: example of the RECORD platform in the context of the ANAEE-France project NOM DE L’AUTEUR JOUR / MOIS / ANNEE H.Raynal (U.MiAT, INRA, Toulouse), A.Chanzy (UMR EMMAH, INRA, Avignon), F. Lafolie (UMR EMMAH, INRA, Avignon), M. El Hadramy (U.MiAT, INRA, Toulouse), D.Maurice ( INRA, Nancy), E. Casellas (U.MiAT, INRA, Toulouse) AgMIP Montpellier 2016

Transcript of Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major...

Page 1: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

Interlinked data and models

using a semantic approach:

example of the RECORD

platform in the context of the

ANAEE-France project

NOM DE L’AUTEUR JOUR / MOIS / ANNEE

H.Raynal (U.MiAT, INRA, Toulouse), A.Chanzy (UMR EMMAH, INRA, Avignon), F. Lafolie (UMR EMMAH, INRA, Avignon), M. El Hadramy (U.MiAT, INRA, Toulouse), D.Maurice ( INRA, Nancy), E. Casellas (U.MiAT, INRA, Toulouse)

AgMIP Montpellier 2016

Page 2: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.02

Outline

• Context• Ontology a tool for interoperability• Different steps

Page 3: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

Context

Page 4: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.04JOUR / MOIS / ANNEE

• Ongoing project (INRA & CNRS)• Aims to develop a national research infrastructure for the study of continental ecosystems

(including agrosystem) and their biodiversity.• This infrastructure brings together modelling platforms and databases for long term

experiments

• Two challenges:• To ease access to and sharing of models and experimental data• To develop interoperability between the different “numerical” elements of the

infrastructure: databases and models

• A lot of tasks in this project. Here is a focus on:Improving interoperability between models (on RECORD platform) and databases using the approach of linked data.

Linked data “a term used to describe a recommended best practice for exposing, sharing, and connecting pieces of data, information, and knowledge on the Semantic Web using URIs and RDF.”(Source Wikipedia)

Analysis and Experimentation on Ecosystems - France

ANAEE-France

Page 5: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.05JOUR / MOIS / ANNEE

Modelling platform• A library of models with different

inputs/outputs (name of variable, units)

• Model typically needs to attach to external sources of data (e.g weather observations, soil characteristics, management practices …)

Why developping « generic » interoperability?

Database• Different sets of experimental data on

agro-ecosystem

• Protocol description • Observed data

Page 6: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.06JOUR / MOIS / ANNEE

Modelling platform• A library of models with different

inputs/outputs (name of variable, units)

• Model typically needs to attach to external sources of data (e.g weather observations, soil characteristics, management practices …)

not a « one shot interoperability »

Why developping « generic » interoperability?

Database• Different sets of experimental data on

agro-ecosystem

• Protocol description • Observed data

Ex. study of the evolution of soil carbon in forest

Modelling platformYasso model (Jari Liski – FinnishEnvironement Institute of Helsinki, 2009 )

SOERE FORETDatabaseLong term experiment in forest

Page 7: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.07

Example of use: design of web application based on

this interoperabilityC stock observed (green triangles) simulated (blue line) - future

Page 8: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

Ontology: a tool for

interoperability

OBOE

Page 9: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.09NOM DE L’AUTEUR / NOM DE LA PRESENTATION JOUR / MOIS / ANNEE

Interoperability: capacity of software components to exchange:• informations• Data• Objects .

Examples:• AGROVOC (FAO) thesaurus to

exchange informations • SANDRE (French administration for

water) Schemadata + languageto exchange data on water

Ontology:• A major tool of semantic web• Community• Semantic• High level of formalization (graph

approach)

A commonly used tool that helps in solving interoperability

problems

Ontology

F. Villa et al. / Environmental Modelling & Software 24 (2009)

Page 10: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.010

• No existing ontology (even by combining different ontologies) to cover all the domaincovered by ANAEE-France

• Domain of application « ecosystem » : wide field decided to choose as startingpoint a generic ontology, and to applied it (rather than an ontology that describesexplicitly the domain)

• OBOE is used for modeling and representing scientific observations that helps in designing frameworks or applications (mainly in the domain of ecological data)

• Analogy can be made between simulation and field experiment, so OBOE can alsobe used in the context of models.

What bring us to the choice of OBOE?

Page 11: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.011JOUR / MOIS / ANNEE

a simple pattern: 5 core classes and 7 properties

OBOE ontology

OBOE conceptual model (Madina et al. 2007)

Page 12: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.012

Example of use of OBOE-ANAEE, in the context of

model & data applied to Carbon in soil forest

the same conceptual schema is applied to describe the observed variable and to describe simulated output variable(same approach for input variables and characteristics of experimental protocol)

Page 13: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.013JOUR / MOIS / ANNEE

International Consortium for Agricultural Systems Applications (ICASA) a naming convention for agricultural model variables. ( AgMIP standards in the harmonized AgMIP Crop Experiment (ACE) database.

OBOE and ICASA Variables

Name (ICASA name) Context level 2 Context level 1 Entity Characteristic Unit

canopy_height crop vegetation cover canopy height

leaf_carbohydrate crop leaf carbohydrate concentrationgrain_carbohydrate_conc grain carbohydrate concentration

growth_stage crop growth stageroot_weight/length crop vegetation cover root mass per length g cm-1growth_respiration_daily crop vegetation cover respiration mass per surface g m-2 day-1dead_canopy_dry_wt crop vegetation cover dead vegetation dry mass per surface kg ha-1grain_dry_weight crop vegetation cover grain dry mass per surface kg ha-2total_biomass_dry_wt crop vegetation cover total vegetation dry mass per surface kg ha-3root_depth crop vegetation cover root length mleaf_area_index crop vegetation cover leaf surface by surface m2.m-2stem_area_index crop vegetation cover stem surface per surface m2.m-2

Atm. radiation surfacic energy global hourly atmosphere

incident shortwave

radiation surfacic energy MJ.m-2

atmosphere air temperature maximum at 10 atmosphere air at 10 cm height temperature KNH4_Nsoil_total soil ammonium nitrogen mass per surface kg ha-1NO3_Nsoil_total soil nitrate nitrogen mass per surface kg ha-1sowing_date sowing date stringploughing_date ploughing date stringsowing_density sowing mass per surface kg ha-1ploughing_depth ploughing depth m

Page 14: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

Different steps

Page 15: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.015NOM DE L’AUTEUR / NOM DE LA PRESENTATION JOUR / MOIS / ANNEE

(Just to keep in mind …) the main underlying concepts and

technologies:

- RDF a general method for describing informations

RDF schema

- SPARQL aRDF query language

- OWL Web Ontology Language

An extension of the web (W3C) with standards that promote

common data formats and exchange protocol on the web

Semantic web:

Page 16: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.016NOM DE L’AUTEUR / NOM DE LA PRESENTATION JOUR / MOIS / ANNEE

• 1rst step: ontology buildingOBOE OBOEE-ANAEE (define all the entities, the characteristics, the relationships …).Online version of the ontology shared by the community

• 2nd step : annotation processAnnotate all the inputs and outputs of each model of RECORDAnnotate all the sets of data of the databases of the infrastructure(in order to have a file describing the inputs, outputs or observations, compliant to the ontology and in RDF format)

• 3rd step: ready for designing application which require data and models. using:

• Distributed architecture. • Request on the different elements of the infrastructure • Exchange (webservices)

Different steps

Page 17: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.017

Example of use: design of web application based on

this interoperabilityC stock observed (green triangles) simulated (blue line) - future

Page 18: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

.018

• Very short introduction to how web semantic can help modellers in the straightforward task : linking database and model.

• A focus on ontology, and OBOE ontology

• Application to ANAEE-France infrastructure

And now …

Ensemble modelling a big work to link models and data. Ontology, websemantic A new perspective/pilot for AGMIP communityensemble modelling (and in particular AgMIP work) ?

Conclusion

Page 19: Interlinked data and models using a semantic approach ... · PDF fileOntology: • A major tool of semantic web • Community • Semantic ... dead_canopy_dry_wt crop vegetation cover

Thank you