Skip to content

From_Chris_M

Jump to bottom
Kai Blumberg edited this page Nov 7, 2022 · 28 revisions

Rapid Development of Knowledge Bases using Prompt Engineering and GitHub CoPilot

using GPT-3 to build ontologies.

Biolink Model – A community driven data model for life sciences

Chris Mungall and Justin Reese at the Ontology Summit 2022 on COVID-19 Knowledge Graphs

Aligning editor/contributor documentation across OBO

Draft of his blogpost

ontotips blog

Creating package-specific term sets for sample metadata standards

PM paper 3

https://microbiomedata.github.io/nmdc-schema/OmicsProcessing/ https://bioregistry.io/registry/biosample http://bioregistry.io/registry/insdc.run

ontolog discussion form Dec 2021

https://groups.google.com/g/ontolog-forum/c/qBROcUuycdI

https://github.com/SDG-InterfaceOntology/sdgio/issues/237 -> Add ontology-term-root annotations to hide BFO terms

https://github.com/BFO-ontology/BFO/issues/221 -> BFON Normalized BFO

Modularization on OBO, with focus on ENVO

yasgui SPARQL Editor and more.

Not necessarily opposed to a shadow, but shadows should be added using a defined DP, using an equivalence axiom with a 1:1 (functional/inverseFunctional) object property, explanation coming in a blog post near you soon....

https://douroucouli.wordpress.com/2019/07/11/proposed-strategy-for-semantics-in-rdf-and-property-graphs/ https://douroucouli.wordpress.com/2019/03/14/biological-knowledge-graph-modeling-design-patterns/ https://douroucouli.wordpress.com/category/knowledge-graphs/ https://github.com/cmungall/owlstar https://kgx.readthedocs.io/

  • An Email regarding OBO chemistry:

Just to be clear, are you using the term 'metaclass' in the sense of a class to which other classes belong, and not in the sense of a super/parent class?

yes, if I would never use it in the other sense

Confusing article here but sort of covers it https://en.wikipedia.org/wiki/Metaclass_(Semantic_Web)

For example, in RDF you define a class using: :class rdf:type owl:Class .

Correct, but of course this disappears at the OWL-DL level

  • SSSOM

Comments welcome on this proposed standard format for ontology mappings: https://github.com/OBOFoundry/SSSOM

This provides a standard TSV and RDF/OWL rendering for mappings. We also have some incredibly rudimentary python tooling (https://sssom-py.readthedocs.io/) and I would like to have robot support (https://github.com/ontodev/robot/issues/312)

I know mappings are not always a priority for us in OBO, we strive for orthogonality, so no need for "mappings" or equivalence pairs between named classes, when we live entirely in the OBO universe. But mappings are incredibly important for broader adoption of ontologies. We don't do a very good job in OBO of providing accurate, complete mappings with provenance and clear semantics, and this is to our detriment. (as someone partially responsible for "xref" I bear much of the blame here)

Also this is a good opportunity to bring semantic standards and to generalize OBO principles to what is often acknowledged to be one of the most painful and ubiquitous problems in bioinformatics.

  • COB units:

https://github.com/OBOFoundry/COB/issues/35

  • Ontology Change Language Data Model

I've created a draft datamodel for a high-level ontology change language: https://github.com/cmungall/ontology-change-language Comments welcome!

This provides a higher-level more user-friendly way of describing changes than either RDF or OWL level diffs; for example, moving a term from one parent to another.

It is designed to be used in a variety of ways

  • generate a high level description of a difference between two versions of an ontology, or two analogous ontologies
  • specify changes to be applied to an ontology
  • a vocabulary for use in summary statistics about ontology changes between releases, similar to GO reports
  • a vocabulary for humans to use in describing ontology changes (it is already linked to GO docs)
  • a spec for higher level operations in tools like Protege (similar to existing diff workflow tool)

It is naturally biased towards a particular abstraction over OWL, e.g. OBO-centric, relational graph centric. It may be possible to accommodate other views.

It's also generic enough it could be used for any Knowledge Graph (whether LPG or RDF) or for ABoxes.

There is no tooling yet, but there is a standard JSON-LD serialization, and JSON-Schema for this, as well as Python dataclasses (that auto-render to / parse from the JSON-LD).

That Chis says is required reading for ontologists.

mixs-demo

Clone this wiki locally