Project Meeting -- 09/18/2019 @ 13:00

[callahantiff]

Meeting Date: September 18, 2019 Topic: Weekly Meeting Attendees: @bill-baumgartner

Proposed Agenda:

• [ ] Comparing full KG to full KG + reasoner inferences
  • Discuss KaBOB script used to convert anonymous nodes to non-anonymous nodes to properly add edges inferred from running reasoners
• [ ] Continue discussing developing KR for clinical to biological concept mappings (see #12)

[callahantiff]

UPDATE:

Proposed Agenda:

• [ ] Comparing full KG to full KG + reasoner inferences
  • Discuss KaBOB script used to convert anonymous nodes to non-anonymous nodes to properly add edges inferred from running reasoners

Before next week's meeting, I need to do the following:

• [x] Verify that all instances have a valid URI and that this information is correctly recorded in a node-level dictionary
• [ ] Verify node and edge count pre- and post-reasoner to ensure that all newly inferred edges are actually being added to the original KG.
• [x] Verify that the assertions ignored by the reasoner are not removed from the KG and thus included in the final version when incorporating the newly inferred assertions.
• [x] Trying running the reasoner without removing disjointness axioms to confirm whether or not those need to be manually removed prior to reasoning.
• [x] If manually removing disjointness axioms, figure out why the numbers before and after removal are not matching up.

• [ ] Continue discussing developing KR for clinical to biological concept mappings (see #12)

• [ ] Before next week's meeting gather more complex examples, those that may be difficult to create constructors from so a strategy can be discussed.

---

@bill-baumgartner - since these are still things I need to address and are likely to be discussed at our next meeting I will leave this issue open. Also, please feel free to edit/comment if there is anything I missed or worth clarifying. 😄

[callahantiff]

@bill-baumgartner - for our meeting tomorrow (10/9/19) I was hoping we could talk about the inaccurate assertions that resulted from running ELK on HPO. The Spreadsheet documenting these issues can be found here.

Goal:

• [ ] Submit an issue to HPO describing our findings

[callahantiff]

Post-Meeting Update

@bill-baumgartner - for our meeting tomorrow (10/9/19) I was hoping we could talk about the inaccurate assertions that resulted from running ELK on HPO. The Spreadsheet documenting these issues can be found here.

Goal:

• [ ] Submit an issue to HPO describing our findings

TODOs:

• [ ] GO back through reasoner results and reorganize to indicate when an error is from the HPO definition or when it is the result of an error in a different ontology that is used in HPO's definition.
• [ ] Identify when reasoner results are not clinically or anatomically correct.
• [ ] Delete all classes except for red hair and skin rash and then run the reasoner. Then, modify each classes' definition to remove red and run reasoner again.

@bill-baumgartner - should I add anything else to this list?

[callahantiff]

Closing this for now and will open new issues for relevant parts of this as they come up through our reasoner evaluation and the v3.0.0 build.