Governance of Ontologies, Vocabularies and Ancillary Research Objects
The governance of ontologies, vocabularies, and other shared resources has been coming up across communities (RDA, EarthCube, ESIP, etc) and across domains (semantic web, software sustainability, documentation, and others).
Topics for discussion include reporting out related to current community efforts, what are the commonalities we can identify for managing and preserving these research objects, what systems are currently being used for this type of governance, what are some of the discussions or efforts for citation of ontologies, and, finally, who is responsible for managing the systems and the objects they contain and what role does (or should) ESIP play in that management.
Strategic plan goals this would fall under: 1, 2, 4
- Ontology governance has come up in during Semantic Web and other groups.
- RDA: There is a metadata interest group that has spun out.
- There is no RDA semantics group because semantics is everywhere.
- WMO started a team investigated terms related to aerosols and gases, however it looks promising.
- Force11 is counting ontology as software using code to manage terms.
- Who is responsible for maintaining these ontologies?
- Often discussed within a group at the project level (i.e. ICD).
- Closest at systematic level might be British Oceanographic Data Center.
- Vocabularies - such as permafrost, sea ice, and snow – have a number of groups working on them.
- Normalized against NVS and NERC, however this is not a domestic service.
- Can submit new terms with definitions.
- Some commitment to persist services.
- Trusted humans curate the ontologies.
- Multiple vocabularies
- For environmental protection, OBO Foundry is a community-curated ontology.
- Ontology has both governance and service components. We should be careful using the portal (i.e. services) to identify governance.
- How does one identify a request for a change to the ontology?
- This does not seem to be done well currently… though github seems to do it best.
- Ontologies could potentially be high jacked, similar to hashtags, through creating false flags.
- Governance of ontologies in not trivial and necessitates broad knowledge.
- There are potentially ways to evaluate ontologies to indicate the level of trust in a flag or ontology as a whole.
- Action Item: development of best practices for creation and maintenance of ontologies.
- Is there a better way to understand changes in GCMD keywords?
- Action Item: potential for a case study using linkable metadata.
- We need to recognize how the data or ontologies will be used and not only identify knowledge.
- ESIP wiki has a semantic tutorial: wiki.esipfed.org/index.php/Semantic_Web_Tutorials
- Ontologies have global scope, software does not. [a response to why Github is not the most effective tool for ontology governance. P.Fox.]
From the Slack (ontology_governance channel)
RDA is all about data sharing , so if it isn't enabling sharing it isn't on the agenda
EarthCube has a semantics working group but due to resource limitations all that is really on the agenda so far is trying to figure out where semantics is happening within EarthCube
Force11 group is working on ontologies as if they are code and are mostly worried about credit
The British Oceanographic Data Center has an ontology governance service (see vocab.nerc.ac.uk)
Terms must have definitions, place in context, reviewed for overlap with existing terms
The BODC could provide the infrastructure for other communities; but they have to provide the curation
What is the issue with OBO Foundary on GitHub?
Need to separate the governance process from the services of any portal
The web protege repository has some of these capabilities
Different types of ontologies need different. types of governance
Semantics committee will host the discussion to create:
- a list of the types of governance needs folks in ESIP have for the vocabularies, ontologies, etc.
- system requirements for a semantics repository
Time to develop best practices for ontologies, metadata, etc. maintenance - for example fixing up dead links in metadata
how do we impact the GCMD keywords, and how they manage them?
proposed project: ISO metadata can encode the URIs to vocabulary terms. These vocabulary terms may be link-able into a semantic web. How can these URIs to vocabulary terms in published metadata become useful or leveraged within the semantic web?
[not 100% sure what the included ISO element references except perhaps a means of referencing vocabulary term IRIs in anchor objects.]
<gmx:Anchor xlink:href="http://linked.rvdata.us/resource/device/100384" xlink:actuate="onRequest">SeaBeam 2112</gmx:Anchor>
<gmx:Anchor xlink:href="http://voc.rvdata.us/device/multibeam" xlink:actuate="onRequest">multibeam</gmx:Anchor>
<gco:CharacterString>multiple formed beam echosounder (mapping sonar)</gco:CharacterString>
Peter Fox suggests it would be helpful to revisit the work already done on the tutorials related to semantics. On the Semantic Web ESIP Wiki:
Group should review the existing ESIP ontology governance process
There is a Semantic Technologies planning meeting Friday at 2pm. We can continue some of this discussion there if there is interest.
thank you all for participating here. if you have thoughts on using slack for conference discussions, let us know in the winter meeting channel. especially in light of the lunch discussion about making telecons more fun. cheers and thanks again!
- development of best practices for creation and maintenance of ontologies.
- development of use cases for ontologies and similar research objects (see joint efforts with RDA Vocabulary IG)
- explore the potential linked metadata project using GCMD keywords and ISO 19115(-2).