'Curification' of additional types of publication URLs

[mbrush]

Outcome of 4-6-23 Data Modeling Call: We debated separating pubs referenced by CURIE from those referenced by URL into separate Attribute objects, to make the UI job of parsing this easier, but for now are keeping a single publications property that can hold a list of curies or urls or free text strings (free text strings would live in a separate publications Attribute object however).

We decided that we will require references to pmids, pmcids, and dois to be reported as CURIES - so when a source provides a URL for one of these types of publications, it should be converted by the KP to CURIE form (using prefixes in the Biolink Model prefix map). This was written into the current supporting publications specification:

URLs for other types of documents /publications will for now be reported as full URLs (e.g. wikipedia articles, dailymed articles, patents, etc).

In the long term, we will want to consider asking KPs to curify other types of document/publication URLs (e.g. wikipedia ruls, dailymed urls). If we decide this has value, we will need a team to look at all types/namespaces of URLs being references as supporting pubs, identify the most common, determine which are possible/useful to curify, and develop guidelines/code to implement the curifications of these. KPs can then use these guides/code to curify these URLs at ingest time - when they are creating the data.

Note that there was some tension about what to ask KPs to do at ingest w.r.t. curification, and what to do in a more centralized way having a service that helps this this that KPs can use could help.

Key contributors for these future conversations: @Genomewide, @cmungall, @edeutsch @mbrush @sierra-moxon . See also comments in https://github.com/NCATSTranslator/ReasonerAPI/issues/410.

[edeutsch]

To be discussed next week at the Relay in the EPC Next Steps session