From the review of #105.
The PR description says PMID "is covered too, which today contributes nothing by accident of the data rather than by design". It is by construction, not by data: build_subsets.rx has no PMID alternative, so no PMID: term is ever extracted and none can reach shared. The not in CITATION filter is correct and future-proof, but the claim that PMID edges exist and are now excluded is not.
Nothing to change in code. Recorded so the next person measuring citation share does not look for PMID edges that cannot exist. If PMID should be extracted, that is a change to rx and to PREF, pinned by the prefix-list tests.
From the review of #105.
The PR description says PMID "is covered too, which today contributes nothing by accident of the data rather than by design". It is by construction, not by data:
build_subsets.rxhas noPMIDalternative, so noPMID:term is ever extracted and none can reachshared. Thenot in CITATIONfilter is correct and future-proof, but the claim that PMID edges exist and are now excluded is not.Nothing to change in code. Recorded so the next person measuring citation share does not look for PMID edges that cannot exist. If PMID should be extracted, that is a change to
rxand toPREF, pinned by the prefix-list tests.