20180220 Ontology Change Improvement Call

Date: 2018.02.06

Attendees: Mike Conlon, Anna Kasprzik, Brian Lowe, Ralph O'Flinn, Muhammad Javed, Kitio Fofack, Christian Hauschke, Juliane Schneider, Violeta Ilik, Andrei Tudor, Damaris Murry

Agenda

• Review vivo.owl
• Development of applicationConfiguration.owl
• Namespaces used in VIVO and Vitro
• What ontologies to list and why?

Notes

Mike was concerned that creating an application ontology file might disrupt the application, but that appears not to be the case. Javed had found a couple of annotation properties that were used in VIVO and Vitro and didn't appear to have definitions. created a file to house the definitions set off discussion of whether it was disruptive to the application answer is no - application is just looking for triples, not ontology declarations went looking for things that are being used as part of the application configuration ontology looking for references to the application configuration namespace in code, files found 13 terms 2 appear to be individuals; rest appear to be properties

Javed Should all 13 be defined in the same place?

Mike Would recommend not separating them out at this time, since that would imply that we should change the namespace and that would affect the code. Should put like with like when organizing the one file: add/edit/delete; qualifiedBy; etc.

Javed Everything works fine in VIVO 1.9? Do we need more testing?

Mike Don't think there is any issue. Appears to have no impact on the running system.
By having JIRA issue and pull request, is pathway in to getting this work into the develop branch and into the distributed code.

vivo.owl discussion:

Mike started this work because I was interested in finding the ontologies that are actually used in VIVO; wanted to write something that could read the code and find namespace references https://goo.gl/vErPWk. (Read distributed RDF; also read code. Not many things hardcoded in code.)

We did find a namespace that was not defined in the distribution (app config ontology). There may be others.
In Vitro it's pretty clean: Vitro has lots of URI references, but seem to be as expected. Some things in the list are not really ontology related, e.g. the Stony Brook namespace for the terminology service table includes node count (subj, pred, or object)

Javed There are few URIs that are used for definition of an ontology. First we should have a JIRA issue to push our organized ontology files to the develop branch. That should go forward For these URIs listed by Mike, divide into ABox and TBox. I've just been working with the TBox. (Action item: Mike will do this; simple to make new headings ABox and TBox.)

Javed will look at TBox URIs. Have two different URIs for two different versions of SKOS. Let's use one of them; update to the latest one. While we're doing that analysis, we should also assess the impact of removing one URI (i.e. from one ontology version). Need to understand if anything will crash. How to inform the community that "you need to do this thing to make it work"; something along those lines.

Mike Agree. Some things we've done we may want to undo. Some ontologies we distribute with VIVO don't really belong in the core. Might mention the OCRE ontologies. Related to clinical study design and seem to be out of scope for VIVO. VIVO needs to be able to say that someone participated in a clinical trial and which one, but the amount of detail for that – I'm not sure that's a VIVO thing.

Christian comes back to point that it needs to be more modular; easier for people to import different ontologies to describe their fields.

Ralph For us, humanities is a big one

Christian For translations, for example, need to keep track of changes so we can translate everything. Don't have time to make diffs of everything; should be tracked automatically if possible.

Javed two different ways of using additional ontologies. Maybe you want to import a specific thing, but maybe you import an ontology that has some overlap with what's already in VIVO. Should we look into issues of mapping to other ontologies?

Violeta think we should look into the mapping

Mike we have a way of expressing things. May be incomplete, and should be made more complete. Other people have other ways of expressing the same things. That's a mapping problem. If the thing is actually the same thing, you can have a sameAs. But I would get concerned if the VIVO community starts referring to the same things with different ontological terms. [General agreement]

Javed If something does not exist in the VIVO ontologies, good: add classes and properties from outside. But if it exists and is modeled in a different way...

Brian will double-check Mike's conclusion that application is unaffected by reorganization.

Mike We know that the ontology list is generated manually, and we know it's not a complete list. We use RDF RDFS, XSD, etc. Question is what is the list, if not a complete list. Two more examples: issue of application ontologies: Vitro public constructs, but don't have other Vitro ontology. there might be four things under the heading of Vitro singletons: have the Event Ontology. We are using exactly one class from the Event Ontology. There are probably half a dozen other ontologies that are in a similar situation but are not on this list. SWO, etc? What kind of a list would be like?

Javed at least list of domain ontologies

Julianne: will make a Google doc of what eagle-i pulls from VIVO

Mike Maybe won't include foundational things. List 8 years in the making; closing in on it. Vitro public (for image-to-individual assertions) versus Vitro ontology (other application-specific entities) How do we feel about more/less of application-type things in the ontology list? clearly should be documented somewhere in the technical documentation, but probably don't need to be in the domain ontology list

Violeta Moved Google docs to Ontology Improvement Task Force folder Think we should make an effort to reach out to communities, e.g. schema.org people. Just have them talk to us and interact with them more.
People like Jim Hendler, Tim Berners-Lee. Use their wisdom from time to time.

Mike Tim Berners-Lee is working with MIT on https://solid.mit.edu/. Have spent a lot of time thinking about permissions and authorities; central thing for social interaction to work on the Web. Like the idea of trying to find related ontological work and reaching out to them. Keep that conversation going in Slack. What projects and ideas we think are important. We do need to be more involved with others.

Christian Sören Auer is getting people together for Open Research Knowledge Graph (ORKG; working paper on Zenodo: https://doi.org/10.5281/zenodo.1157185); move from publication to fact-based publication. Will send white paper around; in Germany in three weeks will be a workshop on this topic.