Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Pitfalls in alignment of observation models resolved using PROV as an upper ontology

AGU Fall Meeting, 2015-12-16

A number of models for observation metadata have been developed in the earth and environmental science communities, including OGC’s Observations and Measurements (O&M), the ecosystems community’s Extensible Observation Ontology (OBOE), the W3C’s Semantic Sensor Network Ontology (SSNO), and the CUAHSI/NSF Observations Data Model v2 (ODM2). In order to combine data formalized in the various models, mappings between these must be developed. In some cases this is straightforward: since ODM2 took O&M as its starting point, their terminology is almost completely aligned. In the eco-informatics world observations are almost never made in isolation of other observations, so OBOE pays particular attention to groupings, with multiple atomic ‘Measurements’ in each oboe:Observation which does not have a result of its own and thus plays a different role to an om:Observation. And while SSN also adopted terminology from O&M, mapping is confounded by the fact that SSNO uses DOLCE as its foundation and places ssn:Observations as ‘Social Objects’ which are explicitly disjoint from ‘Events’, while O&M is formalized as part of the ISO/TC 211 harmonised (UML) model and sees om:Observations as value assignment activities.

Foundational ontologies (such as BFO, GFO, UFO or DOLCE) can provide a framework for alignment, but different upper ontologies can be based in profoundly different world-views and use of incommensurate frameworks can confound rather than help. A potential resolution is provided by comparing recent studies that align SSNO and O&M, respectively, with the PROV ontology. PROV provides just three base classes:
Entity, Activity and Agent. om:Observation is sub-classed
from prov:Activity, while ssn:Observation is sub-classed from prov:Entity. This confirms that, despite the same name, om:Observation and ssn:Observation denote different aspects of the observation process: the observation event, and the record of the observation event, respectively.

Alignment with the simple PROV classes has clarified this issue in a way that had previously proved difficult to resolve. The simple 3-class base model from PROV appears to provide just enough logic to serve as a lightweight upper ontology, particularly for workflow or process-based information.

  • Login to see the comments

Pitfalls in alignment of observation models resolved using PROV as an upper ontology

  1. 1. Pitfalls in alignment of observation models resolved using PROV as an upper ontology Simon Cox | Research Scientist | Environmental Informatics 16 December2015 LAND AND WATER
  2. 2. Overlapping terminology Sources: OGC SensorML OGC Observations and Measurements (O&M)  ISO General Feature Model Semantic Sensor Network Ontology (SSN)  DOLCE UltraLite Biological Collections Ontology (BCO)  Basic Formal Ontology Contentious terms: Observation Process Simon Cox - AGU Fall Meeting 2015 - IN33F-07
  3. 3. SensorML - Process Simon Cox - AGU Fall Meeting 2015 - IN33F-07 All components modeled as processes, including • Hardware - transducers, sensors, platforms • Software Botts & Robin, OGC SensorML – OGC Implementation Specification OGC document 07-000, 12-000
  4. 4. O&M – Process, Observation OM_Observation + phenomenonTime + resultTime + validTime [0..1] + resultQuality [0..*] + parameter [0..*] GF_PropertyType GFI_Feature OM_Process Any +observedProperty 1 0..* +featureOfInterest 1 0..* +procedure1 +result An Observation is an action whose result is an estimate of the value of some property of the feature-of-interest, obtained using a specified procedure Simon Cox - AGU Fall Meeting 2015 - IN33F-07 Cox, OGC Abstract Specification – Topic 20: Observations and Measurements 2.0 ISO 19156:2011 Geographic Information – Observations and measurements ‘Observation’ produces result at a known time Before resultTime: no data After resultTime: data available ‘Process’ is reusable observation procedure
  5. 5. om-lite <http://def.seegrid.csiro.au/ontology/om/om-lite> Simon Cox - AGU Fall Meeting 2015 - IN33F-07 S.J.D. Cox, Ontology for observations and sampling features, with alignments to existing models, Semant. Web J. (2015) Accepted http://www.semantic-web-journal.net/content/ontology-observations-and-sampling-features-alignments-existing-models-0
  6. 6. SSN – Process, Observation Simon Cox - AGU Fall Meeting 2015 - IN33F-07 • Observation, Process both ‘Social Objects’ • Stimulus is the only ‘Event’ M. Compton, P. Barnaghi, L. Bermudez, R. García-Castro, O. Corcho, S.J.D. Cox, et al., The SSN ontology of the W3C semantic sensor network incubator group, Web Semant. Sci. Serv. Agents World Wide Web. 17 (2012) 25–32. doi:10.1016/j.websem.2012.05.003.
  7. 7. Walls RL, Deck J, Guralnick R, Baskauf S, Beaman R, et al. (2014) Semantics in Support of Biodiversity Knowledge Discovery: An Introduction to the Biological Collections Ontology and Related Ontologies. PLoS ONE 9(3): e89606. doi:10.1371/journal.pone.0089606 BCO - ObservingProcess ObservingProcess subClassOf* BFO:Occurrent Simon Cox - AGU Fall Meeting 2015 - IN33F-07
  8. 8. Process-flow model Core PROV Simon Cox - AGU Fall Meeting 2015 - IN33F-07 Developed primarily for datasets, data products, reports T. Lebo, S. Sahoo, D.L. McGuinness, PROV-O: The PROV Ontology, (2013). http://www.w3.org/TR/prov-o/ (accessed February 13, 2014).
  9. 9. Core PROV– aligned with BFO/BCO Simon Cox - AGU Fall Meeting 2015 - IN33F-07 bfo:Occurrent ?? bfo:Continuant bco:ObservingProcess
  10. 10. Core PROV– alignment with O&M Simon Cox - AGU Fall Meeting 2015 - IN33F-07 om:Observation om:Process om:Result
  11. 11. Core PROV– alignment with SSN Simon Cox - AGU Fall Meeting 2015 - IN33F-07 ?? ssn:Sensor ssn:Observation
  12. 12. SSNX aligned with PROV Simon Cox - AGU Fall Meeting 2015 - IN33F-07 M. Compton, D. Corsar, K. Taylor, Sensor Data Provenance: SSNO and PROV-O Together at Last, in: 7th Int. Work. Semant. Sens. Networks, 2014.
  13. 13. Core PROV– alignment with SSNX Simon Cox - AGU Fall Meeting 2015 - IN33F-07 ssnx:ActivityOfSensing ssn:Sensor ssn:Observation Relates to sensor as an asset?
  14. 14. bfo:Continuant Core PROV– all alignments Simon Cox - AGU Fall Meeting 2015 - IN33F-07 ssnx:ActivityOfSensing ssn:Sensor ssn:Observation bfo:Occurrent bco:ObservingProcess om:Observation om:Process Generation of observation data matches a generic process model  PROV is a convenient upper-ontology for alignments Reusable agents
  15. 15. Sampling Features - sam-lite ontology Simon Cox - AGU Fall Meeting 2015 - IN33F-07 S.J.D. Cox, Ontology for observations and sampling features, with alignments to existing models, Semant. Web J. (2015) Accepted http://www.semantic-web-journal.net/content/ontology-observations-and-sampling-features-alignments-existing-models-0
  16. 16. Core PROV– alignment with Specimen prep Simon Cox - AGU Fall Meeting 2015 - IN33F-07 sam:Process sam:Specimen sam:PreparationStep
  17. 17. Specimen preparation and observation trace Lifecycle events modelled as prov:Activity instances • Analysis • Sieving • Grinding • Splitting • Specimen retrieval People and machines modelled as prov:Agent instances • Lab Tech, Geologist • Sieve stack • Mill • Saw • Hammer Simon Cox - AGU Fall Meeting 2015 - IN33F-07 Cox, SJD & Car, NJ Provenance of things - describing geochemistry observation workflows using PROV-O, IN33A-1784
  18. 18. Other alignments and extensions prov:Entity ← :PhysicalEntity ← :Specimen prov:Entity ← prov:Plan ← :SamplingProtocol prov:Agent ← :SampleProcessingSystem ← :GrindingSystem, :PolishingSystem, :DissolvingSystem, :FusingSystem prov:Agent ← :SampleRetrievalSystem ← :FieldSamplingSystem prov:Agent ← :SubSamplingSystem ← :BiasedSplittingSystem ← :SizeSeparationSystem , :DensitySeparationSystem, :MagneticSeparationSystem prov:Agent ← Instrument , Sensor prov:wasAssociatedWith ← :wasControlledBy, :wasSponsoredBy, :wasRequestedBy prov:wasDerivedFrom ← :unbiasedSplitFrom, :biasedSplitFrom prov:wasDerivedFrom ← prov:hadPrimarySource ← :fieldSpecimen Simon Cox - AGU Fall Meeting 2015 - IN33F-07
  19. 19. Summary - in praise of PROV  • Observation models/ontologies use terms “observation” and “process” • Inter-community discussions are vulnerable to misunderstandings • Grounding in traditional ‘upper ontologies’ doesn’t necessarily help! • Generating results of observations is essentially a process-chain  PROV provides a lightweight ‘upper ontology’ that can help Simon Cox - AGU Fall Meeting 2015 - IN33F-07
  20. 20. LAND AND WATER Thank youCSIRO Land and Water Simon Cox Research Scientist t +61 3 9252 6342 e simon.cox@csiro.au w www.csiro.au/people/simon.cox
  21. 21. OBOE observation model Simon Cox - AGU Fall Meeting 2015 - IN33F-07 One Observation is composed of multiple Measurements Each for a different Characteristic of the same Entity
  22. 22. OBOE observation model Simon Cox - AGU Fall Meeting 2015 - IN33F-07
  23. 23. Simon Cox - AGU Fall Meeting 2015 - IN33F-07 om:ObservationCollection  oboe:Observation common feature-of-interest, phenomenonTime om:Observation  oboe:Measurement feature-of-interest, phenomenonTime from collection

×