Author and maintainer identity
Overview
Authors and maintainers on this site are identified by the names they appear under in package DESCRIPTION files. An author comes from the structured author list; a maintainer from the free-text Maintainer field. They have separate pages and are counted separately.
A daily build step attempts to merge spelling variants of the same person into a single profile, erring on the side of keeping people apart.
People & History
Invented people. Ada Okonkwo has a current card while A. Okonkwo appears under "Listed in earlier versions." They are shown as two people because the package page matches former-author cards on spelling alone. A former name is linked only when it has an author page of its own.
Name resolution
Before any comparison, names are lowercased, diacritics are removed, and punctuation is replaced with spaces. Family names are compared exactly (Anderson and Andersson remain separate).
Given names are then compared in five ways:
- Identical tokens
- Jose and José Ferreira. Merges without any identifier.
- Ordered subsequence
- James and James Joseph Balamuta. Requires a shared ORCID.
- Initial expansion
- A. Okonkwo and Ada Okonkwo. Requires a shared ORCID.
- Single-edit distance
- Gerhard and Gerhardt Schwarz. Requires a shared ORCID.
- Name reversal
- Katalin Csillery and Csillery Katalin. Requires a shared ORCID.
Only the identical-tokens case merges on name alone. All other cases require both spellings to carry the same ORCID. A person who recorded an ORCID on one spelling but not the other will remain split.
Clusters larger than five members, or where any pair turns out incompatible, are dissolved back to individual spellings. Organizational names are excluded from resolution.
Scope of name folding
The merged identity is used on author profile pages and career spans. Most other surfaces (the author directory, the directory boards, maintainer pages) use the raw spelling, so a person with two spellings may appear as two separate rows in the directory while having a single merged profile page.
On a merged profile, some figures aggregate across all spellings (package list, reverse dependents, check summary) while others use only the primary spelling (download totals, Bioconductor data).
Maintainer pages group by exact name string with no merging. A change in spelling creates a separate maintainer page.
The asterisk marker
An author profile may carry a superscript asterisk with the note "This page may combine more than one person." This appears when the profile has two or more distinct ORCIDs, suggesting that different people may have been merged.
There is no corresponding marker for the opposite case, where one person is split across multiple pages. A profile without the asterisk is not verified as a single person. The first-time maintainers page notes that a name change can appear as a departure and a fresh arrival.
Profile figures
Key figures on an author profile:
- The sidebar role counts (Maintains, Co-authors, Contributes) are based on the cre, aut, and ctb roles. A person holding multiple roles can appear in more than one count.
- Portfolio Check Status counts individual check-flavor results, not packages. It covers maintained packages only.
- Dependents are the packages that list one of theirs under Depends, Imports or LinkingTo, each counted once across the whole portfolio. Packages that only suggest one of theirs are not counted. The dependency h-index is the largest h for which h of their packages each have at least h dependents.
- Download totals are CRAN only and use the primary spelling.
- "On CRAN since" is the date of the earliest version listing this person, not the package creation date. "Last active" is the most recent release listing them.
ORCIDs are taken from DESCRIPTION as declared and are not independently verified. Author rows exist only for packages with a structured author field. The identity table is rebuilt from scratch on each build with no history, so a new ORCID or spelling change can alter profiles overnight.