Open briri opened 9 months ago
Update the dmphub_modifications array so that it includes a confidence score as well as a note about how the match was determined (e.g. 'matched on grant_id and PI names')
@mariapraetzellis let me know what you think about the following
Here is preliminary logic for generating a confidence score:
Confidence levels:
Related Works (e.g. detected by DataCite EventData, OpenAlex, etc.):
Confidence Match type
--------------------------------------------------------------------------------------------------
Auto Grant ID match (in a scenario where the external system had record of the grant ID)
Auto DMP ID match (in a scenario where the external system had record of the DMP ID)
High 1+ PI ORCID matches, funders match and title/abstract keywords match
High 1+ PI ROR matches, funders match and title/abstract keywords match
High Funders match, repository match and title/abstract keywords match
High 1+ PI names match, repository match, output type match and title/abstract keywords match
Med 1+ PI names match and title/abstract keywords match
Med Funders match and title/abstract keywords match
Med one or more PI names match and title/abstract keywords match
Low 1+ PI names and title/abstract keywords match
Low Funders match, repository match and title/abstract keywords match
Low Funders match and title/abstract keywords match
Grant Information (e.g. detected by NIH Awards API, etc.):
Confidence Match type
--------------------------------------------------------------------------------------------------
High Funder match, opportunity number match and 1+ PI names match
High Funder match, title exact match, 1+ PI names match
High Funder match, 1+ PI names match, project start/end match and title/abstract keywords match
Med Funder match, 1+ PI names match and title/abstract keywords match
Med Funder match, 1+ PI names match and project start/end match
Med Funder match, opportunity number match and title/abstract keywords match
Low Funder match, 1+ PI names match and project start/end match
Low Funder match, project start/end match and title/abstract keywords match
The API functionality is there and being used by the new React UI. We just need to build a schedule-able Lambda that will call them to search for info. Will do that once we've identified a pattern for scanning that reduces the burden on the external API (see issues #66 and #77)
Need to build a fairly generic lambda that can be used to schedule runs for our different API searches (aka ApiScheduler).
Consider having this Lambda fetch DMP IDs that are scan candidates
grant_id
andfunding_status: 'planned'
Then have this Lambda use EventBridge messaging to run scans async to a 2nd Lambda function (aka ApiScanner). This one will perform the following:
funding_status: 'rejected'
grant_id
and changefunding_status: 'granted'
for plansdmproadmap_related_identifiers
All of the above changes should be placed into the
dmphub_modifications
section of the record!