When Hydrospheres Collide

When Hydrospheres CollideLessons in Practical

Environmental Ontologies

John Graybeal, Luis Bermudez

Marine Metadata Interoperability Project

12 October 2006

Marine M

etadata Interoperability Project

MMI: Brief Introduction

• Started 2004 with NSF funding; recent 3-year NSF award

• Mission: Improve the use and understanding of metadata in marine sciences

• International participation and support• Main deliverables: web site, tools, community

– 300-plus members– Numerous open-source tools like VINE, Voc2OWL– Collaborations with many in community

• Technical Lead: Luis Bermudez

Marine M

etadata Interoperability Project

Vocabulary Integration Environment

Marine M

etadata Interoperability Project

Background and Motivation

• Guide a marine data repository to tag data sets with the appropriate data source tag

• Help a data portal discover data through semantic inferencing

• Help an instrument manufacturer to categorize their instruments in a consistent and useful way

• Guide other domains to better categorized their data sources

Marine M

etadata Interoperability Project

Background and Motivation

MMI Workshop Advancing Domain Vocabularies Aug. 2005 Sensor Group

Marine M

etadata Interoperability Project

Plan A: A Sensors Ontology

• Follow-on to Advanced Domain Vocabulary workshop last year– Multiple science domains, plus “sensors”

• Workshop Sensors Team: 6-7 people– Started with GCMD, SWEET vocabularies– Formulated a technique-based hierarchy

• Ontology work was to continue that effort– Some of us were nervous about the work

required to make a sensors ontology

Marine M

etadata Interoperability Project

Plan B: A Platforms Ontology• Good to gain experience with the process• Of direct interest to several activities

– SeaSearch: Roy Lowry– MBARI data systems: John Graybeal et al– Metadata interfaces: Bob Arko

• Easier problem on which to start– Fewer critical concepts for categorization– Fewer existing vocabularies– Fewer stakeholders

• Useful for sensor work later

Marine M

etadata Interoperability Project

Ontology Context

Marine M

etadata Interoperability Project

Tools and Resources

• Concept Schemes: SWEET, CDI Platform Codes, GCMD and Wordnet.

• Dictionaries: Wikipedia, Dictionary.org.• Search Engine: Google, for individual

marine science and technology sites.• Tools: Protégé, SWOOP and Pellet.• Collaboration: WebEx and telephone.• Web Site:

http://marinemetadata.org/sourcesont

Marine M

etadata Interoperability Project

Principal Players

• Luis Bermudez, MMI/MBARI (Lead)• Roy Lowry, BODC• Rob Raskin, SWEET• Robert Arko, LDEO• John Graybeal, MMI/MBARI• Michael Hughes, BODC• Marilyn Drewry, UAH• Kevin O’Neill, BODC

Group Photo

Marine M

etadata Interoperability Project

Principal Customers

• Portals that want to sort or classify data by platform types

• Programmers or Data Managers that want to tag their data sets with a sourceType

• Interoperable systems that want to mediate between two or more controlled vocabularies

• Operators, developers, manufacturers who describe their platforms in metadata records

• Operation managers who manipulate assets • Funders who allocate money to/for assets

Marine M

etadata Interoperability Project

The “Simple” Part

Marine M

etadata Interoperability Project

What Are The “Rules”?• Syntactic Goals

– Short words or phrases– Consistent capitalization/punctuation

• Linguistic Goals– Maintain concept order (noun or modifiers first)– Avoid acronyms – At each level, divide concept space into non-

overlapping concepts that fill the space

• Semantic Goals– Common terms (ideally the most common)– Unambiguous words in English & American– Match other vocabularies where possible

Marine M

etadata Interoperability Project

Class Name Constructs– Adjectives-Noun placement order. In

English adjective goes first. (ResearchVessel instead

of VesselResearch.) Same pattern was applied in

DOLCE, KOALA, PIZZA ontologies.

– Prefer the common marine term over the logic term.

(DriftingBuoy instead of UnmooredBuoy)

– CamelCase preferred vs Hyphen and underscores.

(ResearchVessel instead of Research_Vessel or

Research-Vessel)

(but: Nonautonomous)

The Professor says…The Professor says…

Marine M

etadata Interoperability Project

Little Surprises Everywhere

Marine M

etadata Interoperability Project

Unexpected Meanings

• “AUVs* operate in the hydrosphere”– To me (and many), this is earthbound liquid water– AUVs can operate there, even in canals and ponds

• But to hydrologists everywhere, this includes water vapor– An airborne AUV is not a useful concept– But then, what to do with atmosphere and

terrasphere?

* AUV = autonomous underwater vehicle

Marine M

etadata Interoperability Project

No More Hydrosphere for Us!

Hydrosphere

Marine M

etadata Interoperability Project

What Are The “Rules” (Part 2)?

• How do you organize the ontology? What’s the basis for the framework?– Deployment medium was a clear winner

• How do you choose between 2 equally valid alternatives?– “It depends on how you will use it.”– Helps little in a general purpose ontology– Try to keep model close to reality

• Trouble coming in the sensors work…?

Marine M

etadata Interoperability Project

Organizational Basis

Marine M

etadata Interoperability Project

Criteria to add a new term• It is not already in the ontology.• It can have a property that differentiates

it from its siblings. (e.g. ship and boat. The dimension of a ship is bigger than a boat.)

• A super-class is promoted when similarities are found among concepts. (e.g. Both Buoy and Research Vessel hasEarthRealmBase water. A new class can be created called WaterBasedPlatform.)

• A term can be categorized under 2 or more categories (e.g., Amphibious Crawler).

The Professor says…The Professor says…

Marine M

etadata Interoperability Project

The Results (Ta da!)

Marine M

etadata Interoperability Project

Future Work: Sensors!

Sensor Metadata Interoperability Workshop

October 19-20, 2006

Choosing Standards / Learning Standards

http://marinemetadata.org/workshop06

Marine M

etadata Interoperability Project

MMI: http://marinemetadata.org

Observing Sources Work: http://marinemetadata.org/sourcesont

Ontologies: http://marinemetadata.org/platformonts

Ontology Mailing List: ont@marinemetadata.org

Email: graybeal@mbari.org | bermudez@mbari.org

Thank you