20180529 Ontology Change Improvement Call

2018.05.29

Attendees: Marijane White, Ralph O'Flynn, Brian Lowe, Christian Hauschke, Muhammad Javed, Mike Conlon, Kitio Fofack

Agenda:

How to meet at the conference:

• Christian contacted Eric and Alex about possibly changing the unconference schedule, no results.
• Last meeting it was discussed that we don't want to compete against the other unconference topics.
• What to do now? Schedule something? Javed thinks it will be very difficult to organize something at the conference if we don't try to schedule something in advance.
• Christian says we can't rely on the current published schedule.
• Ralph suggests we define a time shortly before the conference to get together when everyone will be there.
• Mike says we would be able to meet over lunch on Wed/Thurs/Friday, and agrees with Javed that we should schedule something beforehand.
• Christian suggests Wednesday, but Mike suggests that Wednesdays are traditionally smaller attendance, but now that Workshops are free that may no longer be the case, and he prefers to meet sooner than later. Ralph is flexible.
• Decision: meeting on Wednesday at lunchtime.

Documentation:

• Mike has been meaning to ask about this for some time. He is occasionally asked about where the ontology documentation is, and he believes he has an idea of what people are expecting when they ask that question.
• We have the ontology reference in the technical documentation. It has class diagrams, etc. Doesn't describe what an ontology is. No philosophy, no training, etc. It is a reference.
• Marijane wonders if there is another ontology with documentation that sets an example we should follow
• Christian suggests Widoco
• Mike says we often point people at LODE (Marijane thinks Widoco uses LODE)
• Despite this, people ask Mike things like "What can VIVO say about a Person?" He thinks the diagrams are steering people in the right direction, and if you want to see more you can write a SPARQL query. Marijane agrees.
• We could perhaps have more examples, like Book Chapters.
• Marijane suggests setting up a VoCol server, Christian demoed the server he has set up with some German VIVO stuff. Mike wonders if it can show a class hierarchy, in particular a subsumption hierarchy. Christian doesn't think so, but he suggests filing a ticket with the VoCol server.
• Conclusion: we don't need to go out of our way to write more documentation, but we can create more class diagrams as people need them.

Vocabulary/Ontology alignment:

• Something to think about for further conversation at the conference, kind of an open topic.
• For example W3C Time (as asked by Tatiana), W3C Prov (as discussed by Dave Dubin at RDA this year)
• Javed raises a concern about existing data if we would do something like adopt W3C Time in a future VIVO release, Mike says the traditional answer is to provide a CONSTRUCT query to translate data.
  • Brian says it is very simple from the user's perspective, but Javed is concerned that there isn't any documentation about how it works.
  • Mike agrees that as developers of VIVO, we need to understand this, as well as the impact on the community, as was learned in the 1.5->1.6 update. There is a particular impact on people using the API.
  • At the conference last year, both Dave Eichmann and Jim Hendler suggested we could accept both new and old representations, but Mike doesn't quite understand how that could work. Ralph thinks it would be a mess, but Mike says you'd only write to the new one and obscure the fact that there's an old representation in the triplestore, but old queries and tools would still work. Mike agrees it feels messy. Marijane points out it is a very Semantic Web way of thinking about it, and is mostly concerned about whether we'd get some spurious inferences from having both representations in the triplestore. Jim Hendler later told Mike that it is easier in theory than it might be in practice.
  • Decision: Javed suggests that Brian shows the group how the ontology upgrade process works at the Wednesday ontology lunch at the conference, Brian would be happy to do that. He says it is powerful but also simple.

73 open issues in OpenRIF:

• How do we prioritize?
• Mike would like to work on small things that don't impact the code like annotations and additions.
• There are a few hierarchy mistakes in the ontology, Mike thinks fixing these would be relatively low-impact as well. An example is how SKOS is tied into BFO in a way such that the software shows SKOS concepts under Research in the UI. Awards and Honors are also placed in the hierarchy in a way that is surprising in the software UI.
• Javed suggests we try handling five issues in each ontology call.
• Christian says there were 26 issues classified impact:low, should start there.
• Mike says we need to talk to the developers about what branch we should do our work in.
• Javed wonders if we can fork from openRIF. Mike says it's possible, but wants to talk to developers first.
• Marijane says she's supposed to talk to Melissa about OpenRIF and its status, related to Brian's comment a few months ago about abandoning OpenRIF. Biggest concern here is leaving the eagle-i folks without any ontology support. Mike says extraction is still an aspirational goal. Mike says ROBOT did not work for extraction, UF people suggested a tool called OntoPilot.