crossref problems

[robjhyndman]

From Kurt:

Some R Journal publications have their crossref metadata not quite
right.  The one I've recently run into is
<https://journal.r-project.org/articles/RJ-2022-020>, for which

R> x <- rcrossref::cr_cn("10.32614/RJ-2022-020", "rdf-xml")
R> vapply(xml2::xml_find_all(x, "//j.0:creator//j.3:name"), as.character, "")
[1] "<j.3:name>Allison M. Horst</j.3:name>"     
[2] "<j.3:name>Kristen B. Gorman</j.3:name>"   
[3] "<j.3:name>Alison Presmanes Hill</j.3:name>"

looks ok, but

R> vapply(xml2::xml_find_all(x, "//j.0:creator//j.3:familyName"), as.character, "")
[1] "<j.3:familyName>M. Horst</j.3:familyName>"     
[2] "<j.3:familyName>B. Gorman</j.3:familyName>"     
[3] "<j.3:familyName>Presmanes Hill</j.3:familyName>"

R> vapply(xml2::xml_find_all(x, "//j.0:creator//j.3:givenName"), as.character, "")
[1] "<j.3:givenName>Allison</j.3:givenName>"
[2] "<j.3:givenName>Kristen</j.3:givenName>"
[3] "<j.3:givenName>Alison</j.3:givenName>"

shows that the first two have their second given name (initial) end up
in the family name.  Interestingly, the bibtex shown on
<https://journal.r-project.org/articles/RJ-2022-020/#citation> is ok (in
the sense that bibtex::read.bib gets it right ... so something must be
wrong with the code the generates the xml for the crossref registration.

I cannot say how often this occurs: however, e.g. for

  <https://journal.r-project.org/articles/RJ-2022-005/#citation>

clearly even the bibtex is wrong.

I tried the effect of fetching the metadata for all RJ-2022 papers,
using

**********************************************************************
x <- jsonlite::fromJSON("https://journal.r-project.org/articles/articles.json")
p <- basename(x$path)
p <- p[startsWith(p, "RJ-2022")]
x <- rcrossref::cr_cn(paste0("10.32614/", p), "rdf-xml")
names(x) <- p
x <- x[order(names(x))]
one <- function(e) {
    paste(xml2::xml_text(xml2::xml_find_first(e, "./j.3:givenName")),
          xml2::xml_text(xml2::xml_find_first(e, "./j.3:familyName")),
          sep = " @@@ ")
}
fun <- function(e) lapply(xml2::xml_find_all(e, "//j.3:Person"), one)
aut <- lapply(x, fun)
lapply(aut, unlist, use.names = FALSE)
**********************************************************************

which gives the attached output, suggesting that at least for 2022 we
always have the problem with the second given name initial.

[robjhyndman]

More from Kurt:

R> rcrossref::cr_cn("10.32614/rj-2022-013")
[1] " @article{D_Sleire_2022, title={etrm: Energy Trading and Risk Management in R}, volume={14}, ISSN={2073-4859}, url={http://dx.doi.org/10.32614/rj-2022-013}, DOI={10.32614/rj-2022-013}, number={1}, journal={The R Journal}, publisher={The R Foundation}, author={D. Sleire, Anders}, year={2022}, month=jun, pages={320–341} }\n"

which is another example of author names registered incorrectly.  I
could try to fix this programmatically: e.g. the rdf-xml format gives

        <j.3:name>Anders D. Sleire</j.3:name>
        <j.3:familyName>D. Sleire</j.3:familyName>
        <j.3:givenName>Anders</j.3:givenName>

where the name is ok (but the constituents are wrong), but ideally of
course this would be fixed in what is registered on rcrossref.

Two more issues:

* The crossref metadata still have outdated DOI URLs such as the above

    http://dx.doi.org/10.32614/rj-2022-013

  Perhaps simply drop these?  We are not showing them in the BibTeX
  citations on the article web pages either.

* The publisher should really be

    The R Foundation for Statistical Computing