A Marine Platforms Ontology: Experiences and Lessons

1
A Marine Platforms Ontology Experiences and
Lessons
Luis Bermudez (MBARI /MMI) John Graybeal (MBARI
/MMI) Robert Arko (LDEO) Marilyn Drewry
(UAH/ITSC) Roy Lowry (BODC) Kevin ONeill (CCLRC
/NERC) Rob Raskin (NASA/JPL)
Semantic Sensor Networks Workshop , Athens, GA,
Nov 2006
2
MMI - Marine Metadata Interoperability

• NSF starter funding, with SURA(ONR) in-kind
  support, NOAA CSC bridge funds.
• International contributions and support.
• Main deliverables web site, a community,
  demonstrations and tools.
• Goal Enable marine metadata interoperability

3
Background and Motivation
MMI Workshop Advancing Domain Vocabularies Aug.
2005 Sensor Group
4
Started with Platforms.. We thought it was
easier
5
Background and Motivation
Used in www.OOSTethys.org
6
Background and Motivation
MOQuA environment (http//aosn.mbari.org/moqua/)
7
Outcome recent workshop

• Sensor Interoperability Metadata Workshop,
  Portland, October 2006
• MMI should publish controlled vocabularies, in
  particular for technologies that require the use
  of URIs to tag concepts

8
Strategy

• Overall Strategy Public Effort -gt Invited all
  the communities interested.
• One milestone Version 1.0 Beta - May 2006.
• 5 Web conferencing Telecons - 3 hours each.
• Mailing List ont_at_marinemetadata.org.
• One meeting face to face Lunch at Geoinformatics
  Conference, May 2006

9
Ontology - Classes
10
Ontology - Classes
CDI Common Data Index developed by SEA-SEARCH
partners. (30 European Coastal States)
11
Ontology - Classes
12
Ontology - Classes
13
Ontology - Properties of a Platform
20060410
20060514
Extensible Approach
14
Universal Realms - based on SWEET
15
Lessons Learned
16
How is it going to be used ?

• The construction process depends on how will the
  ontology be used.

• Needed URIs to fill in SensorML instances.
• Needed relations between concepts to facilitate
  portal design and discovery of platforms and
  instruments.

17
Need an equilibrium
Domain experts
Tools (e.g. OWL, Protege) experts
18
Ontology guides

• Issues Pizza Guides
• Lack of complex properties such as Mobility.
• Lack of comparison between individuals (Wine
  ont.) vs classes (Pizza Ont.) constructs.
• Ontology expertise in the group is essential to
  provide real-time input.

19
Class Name Constructs

• Adjectives-Noun placement order. In English
  adjective goes first. (ResearchVessel instead of
  VesselResearch). Same pattern was applied in
  DOLCE, KOALA and PIZZA ontologies.
• Prefer the common marine term than the logic
  term. (DriftingBuoy instead of UnmooredBuoy)
• CamelCase preferred vs hyphen and underscores.
  (ResearchVessel instead of Research_Vessel or
  Research-Vessel)

20
Debugging an Ontology

• We use Pellet reasoner to check problems and
  inconsistencies.

21
Conclusions

• More classes were created for WaterBasedPlatforms
  than others, due to the expertise of the
  participants in this domain.
• ObjectProperties are preferred over
  DatatypeProperties.
• Ontology creation is a long process and agreement
  is not easy.
• Keep guides and ontology experts around.
• Keep groups maximum around 5-7.
• Watch out - you may become addicted !

22
Future Work

• Clean up the ontology
• Develop governance strategy.
• Worked with OGC - OOSTethys to make it a part of
  a marine SensorML profile

23
Thank you

• MMI http//marinemetadata.org
• Observing Sources Work /sources
• Ontology /sources/platforms20060508.owl
• Ont List ont_at_marinemetadata.org
• Sensor workshop 2006 /workshop
• Email bermudez_at_mbari.org