Advances in Semantic Harmonization: from the Cryosphere, to the Earth System

Lessons Learned, Tools and Future Directions

https://bit.ly/3iOytPY

Presenters

Agenda

Welcome,Orientation & Overview of Cryo/EnvO/SWEET Harmonization

And a few words about Semantic Harmonization

Lithosphere are the “outer” Earth /planetary layer

No, it’s a Crust and upper part of the mantle.

So continents are part of the Continental Lithosphere?

​

​

Yes, but we Need Thickness as a Qualities Of Layers

​

Harmonizations from the bottom up: Cryosphere vocabulary harmonizations

​

Harmonizations from the bottom up: Cryosphere vocabulary harmonizations

​

Ablation - A graphic illustration of the problem!

​

Ablation Issues - A graphic illustration of the problem!

​

1. Prior to 1980, the term ablation was not used to describe mechanical processes for removal of snow or ice
2. The term is used as
1. a process,
2. a region,
3. an amount of mass removed
3. Ablation is often used as if it were a synonym of surface ablation, when there are other types of ablation (e.g., internal ablation, basal ablation, frontal ablation, etc.) that are important.
4. Some definitions include only snow and ice; while others include "water in any form".
5. Some definitions state that ablation is a process affecting only glaciers, or only glaciers and snow-fields, or only snow cover, or glaciers, floating ice or snow cover.
6. Some definitions include additional discussion of the the physics involved which would be useful for connecting cryospheric ontologies to atmospheric, solar, etc. ontologies.

Imagine what would happen if each community created their semantic resources using only the one source glossary most relevant to them?

Now imagine what would happen later when those communities need to re-use each others data and systems that use those resources?

The Tower of Babel!!!

Using the GCW glossary analysis results to

• Provide definitions for SWEET terms
• To link those SWEET terms to ENVO and other OBO Foundry ontology terms
• To build out the envoPolar subset of ENVO with relevant terms and axioms
• To attribute all of these updates to the people or groups responsible for these changes

So what has the Semantic Harmonization Cluster been doing with Cryospheric terms?

This is not a trivial process!!!

Well actually we haven’t gotten to that term yet! But I note that...

• SWEET does have the term Ablation in the solid phenomenon section (http://sweetontology.net/phenSolid/Ablation)
• ENVO does not have that term but does have the terms
• ice ablation zone
• ice loss process with subclass
1. glacial ice loss with subclass
1. glacial ice ablation process

So what happened to the term Ablation?

Pier - 10 min

​

UN Environment UNESCO

Why it counts and why it needs to scale

from ENVO/SWEET extensions to building global digital ecosystems for the UN Decade of Ocean Science for Sustainable Development and other initiatives

​

What does this mean?

Semantic resources that aspire to be references for the community should be developing MoUs and SOPs around reuse and interoperability at source with expert input - users should not have to hack mappings

​

Only then can we get operational at scale and build globally useful stuff

​

Fancy ideas, but at the end of the day...

We need to agree on what digital resources are about using harmonised semantics for any of the linking technologies to work sensibly

���More focus areas on the horizon...

​

SDG 14 reporting

​

Reporting on marine plastics and debris��UN Working groups now working with ENVO/SDGIO to formulate an official, computable and harmonized classification based on GESAMP and other outputs

Many more harmonisation tasks on the horizon...

We postpone at our own peril … semantic debt has a high interest rate

​

​

Semantic Harmonization

​

​

Practical Principles to

“promote Common Understanding among Earth Science Vocabularies”

Semantic Harmonization:

​

�

​

Lessons learned the hard way:�

• Vocabularies vary significantly in quality and scope
• Hard to access: in PDF, deep in website, etc.
• “Naked terms”-- just a label, no additional description
• Errors-- in spelling, organization/arrangement
• Missing terms, and bits & pieces distributed
• Much redundancy of effort�
• Post hoc vocabulary alignment IS challenging
• Arduous review necessary-- term by term
• Granularity of terms frequently differ (broader, narrower, ambiguous)
• Text definitions/descriptions usually differ (related? how reconcile?)
• Formal axioms always differ (implied, RDF/SKOS/OWL, other) �

Semantic Harmonization:

​

�

​

Some Practical Principles:�

• Vocabularies should be FAIR
• (that’s 4 principles!)�
• Vocabularies should have Governance�
• Re-use, Revise, or Extend--
• before you Re-invent�

​

​

​

Semantic Harmonization:

​

�

​

FAIR:�

​

​

​

​

(Wilkinson et al. https://doi.org/10.1038/sdata.2016.18)

​

...these Principles should apply as well to Vocabularies...

Semantic Harmonization:

​

�

​

�

• Findable

​

• Can Relevant Terms be readily found/discovered?
• Web search?
• ...placement at specific term!
• Do Relevant Terms have a Unique Identifier?
• Disambiguate
• “Re-find” and re-use

​

Semantic Harmonization:

​

�

​

​

• Findable

​

Globally unique identifier for each vocabulary term

Best Practice: dereferenceable HTTP IRI-- provides useful information

�Describe each vocabulary term using recommended fields

terse descriptor-- <rdfs:label> (minimal, insufficient)

detail-- <dc:description>,<rdfs:comment>,<skos:definition>,<IAO:definition>

​

Arrange terms in hierarchies

Best Practice: <rdfs:subClassOf>

​

Semantic Harmonization:

​

�

​

�

• Accessible

​

• Open access, No paywalls�
• Not just hardcopy, or “buried” in PDF�
• Can acquire and/or reference via Web�
• Standard syntaxes, structure

​

Semantic Harmonization:

​

�

​

�

• Interoperable

​

• Built using W3C recommendations

or other well-established schemas�

• Re-use Design Patterns when possible

​

​

Semantic Harmonization:

​

�

​

�

• Reusable

​

• Not just “labels”: when someone looks up a Term, �provide useful information, such as definitions or descriptions
• Include links to related Terms, and indicate how these are related
• Terms should have unambiguous, globally-unique identifiers
• Dereferenceable HTTP URI’s

​

FAIR Vocabularies:

​

�

​

rdfs:label

​

HTTP URI

rdfs:subClassOf

IAO: definition

rdfs:comment

Extra credit axioms!

Semantic Harmonization:

​

�

​

�

• Vocabularies (like Nations!) need Governance

​

• Community development and vetting
• Mechanisms for maintaining access to the vocabulary, and extending/revising it
• Long-term sustainability (ESIP can help?! COR)
• PIRIs
• Operations
• Open, collaborative
• Support?

​

Semantic Harmonization:

​

�

​

�

• Re-Use, Extend, and Revise not Re-Invent!

​

• Established Earth Science vocabularies exist, e.g.
• SWEET, EnvO, GCMD, CF Metadata, SVO,more...
• Modernize and harmonize
• Extend and revise

​

• Established General vocabularies exist, e.g.
• Schema.org, Dublin Core, DCAT2, VIVO, DDI, more…
• Re-Use!

​

​

​

Re-use

Extend

Revise

Semantic Harmonization:

​

�

​

​

If Vocabularies become FAIR �

• Easier to align:�
• owl:equivalentClass
• owl:sameAs
• skos:related

​

• Interoperable “out of the box”
• Pick needed terms from multiple vocabularies

​

​

​

Re-use

Extend

Revise

Semantic Harmonization:

​

�

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

OUR CHOICES

?

Example Tooling

• New ENVO collaborative workflow using OBO tools (Robot)
• Guides users through the process of creating new terms
• Intended for non-ontologists!

​

• Overview:
• Organize new term requests in standardized google sheet template
• Step by step guidance to fill out the appropriate columns
• Protocol for ontologists to add request to ENVO (not covered in presentation)

ENVO-Robot-template-and-merge-workflow

Example template

Kai Blumberg Univeristy of Arizona & BCODMO kblumberg@email.arizona.edu

Soil Ontology and Informatics ESIP Cluster

Our mission is to connect the soils research community, and those whose research incorporates soils data, to informatics tools for better research.

​

We will do this by 1) promoting research exchange between the informatics and soil science community, 2) creating how-to resources and tools for soil data harmonization, and 3) targeted development of semantic resources such as ontologies and other informatics tools for soil science.

​

-2020 ESIP Soil Ontology and Informatics Clustering proposal

Kathe Todd-Brown (UFlorida) ktoddbrown@ufl.edu

Soil data is decentralized with heterogeneous structure, but sometimes related.

NRCS KSSL/SSURGO

(survey)

ISCN

(organic carbon)

ISRaD

(radiocarbon and fraction)

SIDb

(incubation)

VanGestel

(field experiments)

Crowther

(field warming)

CC-RCN

(Coastal soils)

SoilHealthDB

(health)

SoDaH

(long observations)

Ameriflux

(ecosystem flux)

WoSIS

(survey)

Kathe Todd-Brown (UFlorida) ktoddbrown@ufl.edu

Soil Ontology and Informatics ESIP Cluster

Past wins: Review of soil data products https://esipfed.github.io/soil_data_model_survey/

Current goal: Developing a harmonization tool (ontology/knowledge-graph/thesaurus) for soil biogeochemical data

Future hopes: Implementation in an open-community soil database https://github.com/ISCN/SOCDRaHR2

​

Come join us! https://wiki.esipfed.org/Soil_Ontologies_and_Informatics

​

Kathe Todd-Brown (UFlorida) ktoddbrown@ufl.edu

Agriculture and Climate Cluster Report Out

Brian Wee

https://youtu.be/udTrF6CLKdU

​

Thanks!

ESIP Semantic Harmonization Cluster

Mailing List: https://lists.esipfed.org/mailman/listinfo/esip-semanticharmonization

Wiki Page: https://wiki.esipfed.org/SemanticHarmonization

​

52

Join us:

Telecons via GoToMeeting: https://global.gotomeeting.com/join/800617765

• Third Wednesday of the month at 2pm EST

​