OPAL project and ontologies in astronomy

An ontology portal for astronomy, planetary Sciences and heliophysics: towards semantic interoperability B Cecconi (1), C Jonquet (2), E David (1), L Fretel (1), L Debisschop (1), I Bourouche (2), S Aicard (1). PADC Seminar, Paris, Dec. 2025 (1) Observatoire de Paris-PSL, Meudon, France! (2) INRAE, Montpellier, France! FAIR-IMPACT: EU funded project (Grant Agreement: 101057344). https://fair-impact.eu ! OPAL: funded by OSCARS, an EU funded project (Grant Agreement: 101129751). https://oscars-project.eu/ OPAL: Ontology Portal for Astronomy Linked-Data What is an Ontology (still learning) •An ontology is a structured description of knowledge! •Depending on the level of description: terminology, vocabulary, thesaurus…! •Depending on the content: schema, information model…! •A schema defines classes and their properties! •An ontology can also include concepts and instances. ! •Wider term: Semantic Artefact! ! Note: in the astronomy IVOA, schema == data models Predicate Subject Object What is Linked Data (still learning) •goes back to the roots of the web! •linking resources using oriented# and tagged graph! •Triples with subject > predicate > object ! •RDF (Resource Description Framework) is W3C standard! •Several serialisation (RDF/XML, Turtle, JSON-LD…)! •SPARQL is the query language! •Many schemas available (e.g., SKOS, DublinCore, # Schema.org…) Vocabularies and semantics Why would we need this in the first place •Agreement on nomenclature and definitions: required for understanding # At programmatic level: require for interoperability! •Simple example 1: search for dataset acquired by the Hubble Space Telescope. # Some data providers use HST, other Hubble Space Telescope, or just Hubble. ! => a community agreed terminology can simplify this step.! •Structured metadata requires definitions of schemas/data-models, list of allowed values for programmatic access… ! •For interdisciplinary studies, mappings between vocabularies and schemas are often necessary! •Simple example 2: search for data with subject “planet” in an astronomy repository (e.g., # ESO, or CDS) or a solar system repository (e.g., VESPA or ESA/PSA). # Astronomy repository assumes “planet”=“exoplanet”, solar system repositories usually ! have both “planet" and “exoplanet” terms. Semantic Artefacts in Astronomy interoperability, subjects, schemas •Astronomy: # - IVOA vocabularies = many controlled lists for interoperability (RDF)# - UAT thesaurus = subject keywords for publications (RDF)# - IVOA metadata schemas = data model for interfaces, protocols, formats, registry… # (XML schemas only)! •Heliophysics:# - metadata schema for the resource registry (XML schema only)# => list of terms embedded in schema # - UAT thesaurus = subject keywords for publications (RDF)# - New projects on semantics growing up.! •Planetary Sciences:# - information model for the archive (OAIS implementation, XML schema)# => list of terms embedded in schema# https://ivoa.net https://ihdea.org https://planetarydata.org How does this feed into Open Science? Linked-data and publications •Use case: When publishing data, the dataset landing page contains linked-data elements (often in JSON-LD following schema.org).! •Many ways to describe additional metadata on a research product:# - attribution (e.g., with schema.org or datacite)# - provenance (e.g., with Prov-O or VO-Prov)# - access (e.g., with DCAT or Schema.org)# - content (e.g., with SoSO (science on Schema.org), or I-Adopt)# etc.! •All those metadata schema and content can be represented in RDF # (in XML, TTL or JSON-LD), and can be fed into graph databases. # => enable powerful search capabilities. Theory and Modelling Defining parameters precisely •IVOA VOTheory group => vocabularies# for astrophysical modelling# # still TBD for planetary and heliophysics ! •Connecting data (observed quantities) with# numerical modelling inputs or outputs# # I-ADOPT framework under study MDL Issues with Ontologies Wide world, many sources •spread out! •in different formats, of different size! •with different structures! •increasing number! •overlapping Number of ontologies in BioPortal UAT AstrObject IVOA Object Type EPN Target Types Overlapping semantic artefacts OntoPortal a Semantic Artefact Catalogue •OntoPortal: Developed by Stanford: https://ontoportal.org# More recent distribution (with new capabilities) by INRAE (France): # https://agroportal.lirmm.fr ! •OntoPortal Alliance — international coordination# Yearly meeting (next one: Sept 2026, Paris, France)! •Impulse with FAIR-IMPACT (Horizon-Europe) project# Coordination between portals avoid replications of semantic artefacts.# # # # # OPAL - Ontology Portal for Astronomy Linked-data Status: Vocabularies currently being worked on •UCD (Elsa) - Rewrite as an IVOA vocabulary. Update UCD processes. ! •ProcessingLevel (Elsa): # - many lists of terms => propose vocabulary covering all cases, and listing the labels used in the various contexts. ! •UAT (Unified Astronomy Thesaurus) extension for heliophysics (Elsa):# - led by Ryan McGranagham (NASA/Heliophysics). # - reorganised hierarchy in UAT for heliophysics terms # - new terms for heliophysics regions, new top level concept (Heliophysics)# # Starting discussions: similar work for planetary sciences. URIs, Metadata and Provenance How could we improve vocabulary quality. •Vocabulary metadata # - possible include metadata for, e.g., preferred namespace prefix, version, vocabulary URI, version vocabulary URI# - MOD (Metadata for Ontology Description and Publication Ontology) ontology proposes generic keywords [developed within FAIR-IMPACT]! •IVOA VEP info stored within vocabulary! - currently, VEP information is disconnected from vocabulary# - most of the VEP sections could be included in the published# vocabularies, using MOD properties. # - MOD includes, e.g., a “knowUsage” property equivalent to “used-in” # - Currently testing URIs, Metadata and Provenance How could we improve vocabulary quality. •Vocabulary and Term URIs # - current all IVOA vocabs are http://ivoa.net/rdf/… # - many (many!) communities uses https://w3id.org (with redirections from a community managed git repo: https://github.com/perma-id/w3id.org)# - bad experience with purl, but this community seems organised to be more resilient. Next steps New vocabularies & Mapping between vocabularies •Heliophysics (IHDEA group) willing to adopt OntoPortal-Astro.eu for their new ontologies. ! •Work with IHDEA to produce SPASE RDF schema/ontology! •Work with IPDA to produce PDS4 RDF schema/ontology! •Work with VAMDC (virtual atomic and molecular data center) to produce an ontology from their vocabulary! •Ontology for Instrument-types ? (based on PDS4 information model and archive content)! •Vocabulary for Instruments (mounted on ObsFacilities): PDS4 context products; # SPASE instruments, NSSDC instruments…! •Start building domain ontologies, e.g., Small bodies 🎉 Next steps Vocabulary development and management •Propose new mappings using OntoPortal between ontologies/vocabularies.! •Improved FAIR practices in vocabularies# - e.g., attribution, provenance, versioning, licence…! •Update practices to include each term’s provenance in ontology! •Identify related semantic artefacts in other portals (e.g., in EarthPortal: GEMET, SOSA…)! •Work with data alliances (IVOA, IHDEA, IPDA) to produce semantic # artefacts from their XML schema.! •Work with data alliances (IVOA, IHDEA, IPDA) to use OntoPortal-Astro Next steps Knowledge graphs & representing variables •EU project (started in Feb 2024): OSTrails, with focus on building # Science Knowledge Graphs (SKG) => astronomy pilot project # led by ObsParis ! •Participation to SKG RDA working group ! •Explore use of RDA I-Adopt framework (https://i-adopt.github.io) # for describing measurements and quantities and link to concepts # in our ontologies.! •Explore Science-on-Schema.org (SOSO) to expose metadata. # Following up work done by R. Ringuette (IHDEA/Heliophysics) on SPASE registry (mapping SPASE to schema.org, for richer landing page metadata) * Ringuette et al. (2025). Improving FAIR for HDRL’s Metadata: Mapping to # Schema.org and DataCite. doi:10.5281/zenodo.15594513.