20170717 Ontology Change Improvement Call

Date: July 17, 2017

Attendees: Mike Conlon, Marijane White, Juliane Schneider, Brian Lowe, Damaris Murry, Graham Triggs, Linda Rowan, Plato Smith, Muhammed Javed

Agenda: https://wiki.duraspace.org/display/VIVO/2017-07-17+Ontology+Improvement+Meeting

Mike: Anna needs to meet an hour earlier. That is getting kind of rough for Marijane.

Marijane: Literally got out of bed 20 minutes ago. But I could try to get up earlier.

Mike: Brian would probably also like an earlier meeting

Brian: Yes. This is 6pm my time.

Mike: any other announcements?

Marijane: Dave Eichmann should have ontology usage and extensions all collected and crunched by now. Will talk to him tomorrow.

Mike: W3C time ontology is now a candidate. Starts out similar to what VIVO has, but then gets more complicated.

Marijane: Do they do the begin/end intervals like the CRM?

Mike:

We make a distinction between a self-reported ORCiD and those that ORCiD confirms and adds to the triplestore.

Review the ontology change process: https://docs.google.com/document/d/1FVxKmYHfDglgrBHseKcssBZ5NUXMgQ9lBJVWyNpgpdA/edit#

Added some commments to this doc.

Jump down to actions. So there is this issue of creating a canonical vivo-isf-ontology that the VIVO ontology can be extracted from.

If you extract the core, you only get the core. No bibo, etc. You do get the OWL types, and you get the VITRO assertions for the UI.

You have to be a site administrator on a VIVO instance to export the ontologies. Probably have a recent set that could be placed in the Google folder.

Marijane: And we might have to put some stuff back that was removed. Mike, you weren't here the last time we met, but we discovered that some Person subtyptes are missing from the ISF because the idea was that they'd be transitioned to roles, but it seems that happened on the ontology side but not the software side.

Javed joins the call

Mike catches Javed up on what's been discussed so far

Mike: Discovered that VIVO does not export individuals.

Javed: how did you export the file you sent to me?

Mike: Just

Marijane: So what we need is an export that gets everything that VIVO uses, and not just core.

Mike: Let's discuss the principles.

Marijane: I like that "don't break things" is the first principle.

Javed: Concerned that old ontology will not work with new ontology.

Marijane: I realize that putting everything back in might not work. But maybe we put everything back in the ISF and extract the various released versions. Maybe a version that would allow people to upgrade?

Javed: so what we need is to communicate.

Mike: That is what I am getting at with the second principle.

Mike and Javed discuss types changes

Mike: ontology changes and software changes need to go hand-in-hand.

Let's talk about process.

Marijane: When we first put this together, we were cribbing off of the eagle-i process.

Mike: There's a lot of eagle-i stuff in the repo. Need a VIVO tag.

How does eagle-i track status of issues?

Marijane: I don't think eagle-i has a set of status tags. Maybe Juliane knows?

Juliane: I don't think there is. Tenille was just making issues.

Marijane: so it sounds like we need those tags. Plus an eagle-i tag for the eagle-i specific stuff.

Mike: Tags need to include things like "awaiting public comment". The

Needs to be some kind of analysis.

Marijane: in terms of the impact of the change?

Mike: Yes. additive changes, things that will change database contents, things that will require software changes, etc.

Marijane: Do we want to proactively assign tags in need of analysis to people who have the expertise, or assign a tag like "needs technical review".

Mike: If we have a team of reviewers and committers, I am fine with assigning tickets.

Javed agrees

Marijane: So it sounds like people are in favor of assigning issues for review. So if we need to asses, say, whether a change to the software will be needed, we can assign it to Graham.

Mike: So we have this vivo-isf-ontology repo, I created new ontology items in the community repo.

Marijane: Ontology changes vs community requests. Need to clean these up. So do we need an enduring triage process?

Mike: Yes

Marijane: I will find out if you can easily move issues between github repos

Mike: We have to also track them in JIRA, would like to only have two places to do this, not three.

Need to distinguish between ontology issues and ontology support issues, like not exporting individuals.

I will be the secretary for JIRA issues. Issues in JIRA and Github will link to each other.

And then there's an editing step. Eagle-i process is very terminology specific, need to extract these instructions and edit them.

Javed: So the plan is to do all the editing in vivo-isf?

Mike: Yes.

discussion of details of edit and commit process, updates to google doc

Javed: comment on changelog. Some kind of structured format that can be machine-processed.

Marijane: makes me wonder if there's a changelog ontology out there.

Javed: I did that as a part of my PhD. Graphs to record ontology changes, etc.

Mike: I think we need a presentation on this.

Javed: we can talk about it at the conference

Mike: but how about at a future instance of this meeting?

Javed: ok, I can put something together.

Brian: What we used to do with the early versions of the ontology is that we produced a tabbed text file -- used to be a Protege plugin that produced a table of diffs. Data migration used to run off those tabbed text files. So there is a precedent for what Javed wants to do.

Javed: There are different ways to record it. The differences between versions, or log of changes applied. I will give a presentation on this at a future meeting.

Mike: we should explore what Web Protege supports for diffs.

Can we talk extraction? Don't have a lot of time left.

Marijane: two ways, Shahim's OwlCL tool or the OBO ROBOT tool. The latter is actively supported, Melissa recommends.

Mike: The Stanford folks use ROBOT for some pretty big projects.

Would like it if ROBOT worked like the VIVO export, just by namespace.

Marijane: What if the ISF has stuff in namespaces that isn't currently part of VIVO?

Mike: that's fine, we want to see the difference between the two.

Marijane: this seems like a good place to wrap up.