20180726 Ontology Change Improvement Call

2018.07.26

Attendees: Violeta Ilik, Marijane White, Ralph O'Flinn, Mike Conlon, Muhammad Javed, Juliane Schneider

Agenda: https://wiki.duraspace.org/display/VIVO/2018-07-26+Ontology+Improvement+Call

Congratulate Violeta for being on the VIVO leadership group!

Extracting VIVO from VIVO-ISF:

• Mark from Stanford recommended Mike talk to Linda Shapiro at University of Washington.
• We should talk to the ROBOT folks about how to make it work for us

Brian Lowe showed us (at the conference) how to set up scripts for triplestore updates in the event of ontology changes:

• all the technology exists in VIVO to make changes, adding and removing triples, mostly with SPARQL CONSTRUCT query
• VIVO actually checks those scripts every time it starts! The database comes with a triple that asserts the ontology's version, at least since 1.6.
• Ralph warns us against removing the version triple. Bad things happened.
• Triple currently asserts version 1.7, which is neither the version of the ontology or the software.
• Mike is working with Brian Lowe to document this.
• Mike thinks the version asserted by the triplestore should be the version asserted by the ontology, however, the software could be incompatible with the software.
• If we make changes, we are the people who will write the SPARQL queries

Ontology Versioning:

• Christian suggests that we clarify our versioning
• Now is a good time to do it, because we have vivo.owl
• We may need to standardize on 1.7, because that's what the software thinks it is. In which case, the ontology itself should have an assertion that says this.
• Mike may open a pull request to do this
• The MIRO article makes some recommendations about assertions about the ontology itself. Mike would like to review that to see if we can meet some of those recommendations; expects it will say something about assertions the ontology makes about itself.

Ontology subsumption hierarchy viewer:

• Mike wrote something in d3 to read vivo.owl and show subsumption hierarchies and shared his screen to show us. Colors indicate namespace.
• In a BFO world we might expect everything to roll up to bfo:entity
• We have a bunch of stuff that is not connected, mostly things from other ontologies, like vcard.
• Some things were made Concepts that aren't actually
• Javed asks what it means when we see a subsumption error? Why is Mike expecting a superclass?
• Mike's thinking is that in BFO/realist worldview, you're supposed to know what the things in your ontology are
• Javed wonders why we need everything to subsume up to BFO?
• Marijane wonders why not everything is under bfo:entity? Because the team working on CTSA connect ran out of time.
• Not everyone uses BFO. Biomedicine and military spaces use it, but scholarship does not. So if we want to use some of these ontologies, we need to figure out where they fit.
• The upper level ontology is supposed to shape your thinking
• In talking to the UF ontologists, Mike has learned there are some things unique about VIVO that make it challenging to work with. The number of properties is unusual.
• VIVO is heavily influenced by record-based thinking. Not used to thinking about what the real things actually are.
• The decision to reuse a lot of other ontologies also makes things more challenging. For example, is BIBO dead? Should we be in contact with all the ontologies we're using? Should we try taking over ontologies that have died?
• Need to evaluate status of all the ontologies we're using. Possibly replace terms from dead ontologies? Replace obsolete concepts?
• Marijane notes that the BIBO website is working today, and that they appear to have a Google Group and a GitHub repo
• Mike may start making pull requests to put things into BFO.

Priorities for the ontology:

• Cleaning -- Mike found a redundant class assertion with ROBOT -- vivo:EmeritusFaculty has a redundant assertion that it's a foaf:Person
• Adding new terms, like humanities. There are several ontologies available, in addition to the Duke ontology, which illustrates the gap we're trying to fill, and you can see all of them at their VIVO instance.
  • Mike thinks we should research various humanities ontologies
  • Let's make a document in our Google Drive folder to work on this
• Modules, especially clinical trials, OCRe, etc
  • https://deffopera.dk is working on humanities ontologies