Notes: SKG Graph Review

https://confluence.oraclecorp.com/confluence/display/IBS/Guide+-+Contributor+guide+to+SKG CLOSE 1st 4 subtasks in EPIC - so it pops up on their board https://devops.oci.oraclecorp.com/devops-coderepository/namespaces/axuxirvibvvo/projects/ODA/repositories/skg-data-model-review?_ctx=us-phoenix-1%2Cdevops_scm_central https://confluence.oraclecorp.com/confluence/display/IBS/Healthcare+Knowledge+Platform+Intake https://oracle-one.slack.com/archives/C07SWLPMMA6/p1773417600311439 Image 1 (Jira): https://jira.oraclecorp.com/jira/browse/CDAF-120897 Image 2 (DevOps Code Repository): https://devops.oci.oraclecorp.com/devops-coderepository/namespaces/axuxirvibvvo/projects/ODA/repositories/skg-data-model-review?_ctx=us-phoenix-1%2Cdevops_scm_central Graph Reviews Codex Wizard: Graph Reviews README - explains PR to the repository Intake JIRA CDAF-### Pull Request SUMMARY OF MEETING: New Review Process (What Changed) Manual graph reviews by the SKG team are no longer the primary path - too unscalable. Reviews now go through an automated "Codex Wizard" first, then a human review after. Human reviews happen asynchronously in recurring Wednesday/Thursday meetings - no scheduling required. How to Submit a Graph Review Follow the README in the Codex Wizard repository (link posted in Slack/Confluence intake page). Upload your trig file (model) and sample SPARQL queries to the wizard. Resolve all feedback the wizard raises (deterministic Python checks + best-practice markdown files). Open a PR to the Codex repository. Close the first four subtasks in your intake Epic - this surfaces your review on the team's dashboard. When You Need Another Review Minor schema changes: use your judgment, likely no new review needed. Significant/moderate changes: go through the review again. No formal measurement system exists - you are trusted to self-assess. You can run the wizard tools anytime outside the formal process to inspect your changes - takes under 10 minutes. Escalation If your PR is on the board and you need faster turnaround than Wed/Thu, post in the website support Slack channel and request escalation. POC vs. Higher Environments No review required to upload or iterate in POC - iterate freely. Review is required only for promotion to staging/production. --- FROM OFFICE HOURS SCHEDULED MEETING:SKG Office Hours - Intake Updates Wednesday, March 18, 2026 9:00 AM-9:30 AM Open office hours for the below announcement Announcement We are releasing a codex based intake wizard and accompanying process updates to the SKG intake process. Details below. Are you affected? If you are submitting an SKG Intake request and are not sure what category it falls into, please use the wizard based flow to help you identify the proper category. The categories are: 1. New Federated Data - This is when you as an engineering/science team intend to contribute and manage data yourself with SKG. This data could be part of the core graph that distributes to all customers or client specific data (examples include medical necessity guidelines, payer/provider contracts, and Carepathways) 2. New SKG Managed Data - This is when you are requesting common vocabulary data to be added to SKG that will be owned and maintained by the SKG/ECF teams. Examples include LOINC, CPT4, ICD-10-CM/PCS, etc... 3. New Functionality - You require SKG to support net new platform functionality to accommodate your use case. If you already know the category and it falls under #2 or #3 noted above, please continue to follow the process in our wiki here If you are contributing federated data, please read on. Details Review the ReadMe in our wizard repo for full details At a high level the process is as follows: 1. Develop your data model and SPARQL queries in accordance with our best practices already documented in our intake wiki and review checklists, including the creation of an intake request as is currently done in the intake process. 2. Once you are ready for a review, create a new branch off of the # skg-data-model-review repo with the jira ID of your intake as the branch name. Follow the instructions in the Readme to go through the wizard and submit your data model, sample SPARQL queries, and any other related documentation to the wizard. 3. You will receive feedback from the wizard that includes required and recommended changes. If you disagree with any suggestions, denote this in the review file that is produced. 4. Submit a PR when you are ready for the final human review by the SKG team. The team will complete a final review and approve the PR for merging once the data model is approved. 5. Your merging of the branch will be the signal that the data model is considered approved for promotion to higher environments. Logistics This process will be in place for any graph reviews that have not started as of Monday March 16. If you are already in the process, you are still welcome to go through the codex wizard and get automated feedback, though you are not required to do so. We will be iteratively feeding our manual reviews back into the wizard so it will improve over time. We welcome any feedback on the above process. I will be hosting 2 special office hours (the existing mailing groups will be invited to these office hours) next week to answer any detailed questions about the process changes. In the meantime, please feel free to drop questions in thread.