Tracking how a literature changes between retrievals

library(scopusflow)

Run the same search a few months apart and the result will have grown, and perhaps lost a record that was re-indexed. This article shows how to see what changed and how to merge retrievals without duplicates. It runs offline on the bundled example_records, a corpus of 138 real journal articles that stands in for ‘Scopus’ records, since those may not be redistributed. That corpus is a complete harvest of one query from 2015 to 2024, so a pull that stopped at 2023 and a later one that reaches 2024 are both slices of the same search.

The baseline

The first retrieval ran at the end of 2023 and returned everything published up to then.

baseline <- example_records[example_records$year <= 2023, ]
nrow(baseline)
[1] 124

A later retrieval

A year on, the search is repeated. It now picks up the 2024 papers, and one record that was present the first time has since been re-indexed and no longer matches.

later <- example_records[-1, ]
nrow(later)
[1] 137

What changed

scopus_diff_dois() reports which DOIs were added, removed or unchanged between the two retrievals, and prints the counts in each category.

changes <- scopus_diff_dois(old = baseline, new = later)
print(changes)
<scopus_doi_diff> 14 added, 1 removed, 112 unchanged, 11 earlier and 11 later
records without a DOI not compared
# A tibble: 127 × 2
   doi                            status
   <chr>                          <fct> 
 1 10.1002/adfm.202315137         added 
 2 10.1002/asia.202400548         added 
 3 10.1002/slct.202302535         added 
 4 10.1016/j.cej.2024.148822      added 
 5 10.1016/j.diamond.2024.110842  added 
 6 10.1016/j.isci.2024.111696     added 
 7 10.1016/j.jallcom.2024.175000  added 
 8 10.1016/j.jallcom.2024.177248  added 
 9 10.1016/j.jpowsour.2024.234127 added 
10 10.1016/j.jpowsour.2024.236149 added 
# ℹ 117 more rows

The newly indexed papers come back as added, the records present both times as unchanged and anything dropped from the later pull as removed. The counts work out at fourteen added, one removed and 112 unchanged. Fourteen are added because that is how many of the 2024 papers carry a DOI, and the re-indexed record takes the unchanged count from 113 down to 112. Records without a DOI cannot be tracked this way. There are eleven in each pull here, and the printed line counts them, as do the attributes n_old_without_doi and n_new_without_doi.

To act on one category, filter the table, which is an ordinary tibble.

head(changes[changes$status == "added", ])
doi status
10.1002/adfm.202315137 added
10.1002/asia.202400548 added
10.1002/slct.202302535 added
10.1016/j.cej.2024.148822 added
10.1016/j.diamond.2024.110842 added
10.1016/j.isci.2024.111696 added

Comparing records

scopus_diff_records() compares the records themselves, matching them as scopus_combine(dedupe = TRUE) de-duplicates them: on a shared ‘Scopus’ identifier, or on a shared DOI where one of the two records lacks an identifier. A record without a DOI is then tracked by its identifier. Where a record is found in one pull only and the other pull gives its DOI to a different record, it is marked conflict. It may have been re-indexed under a new identifier or may be a distinct paper sharing the DOI, and which it is can be decided only by inspecting the records.

The bundled records carry no identifier, so here they are matched on their DOIs, and the eleven without one in each pull cannot be compared. Each row carries the record’s title and year and its citation counts in both pulls.

records_changed <- scopus_diff_records(old = baseline, new = later)
print(records_changed)
<scopus_record_diff> 14 added, 1 removed, 0 in conflict, 112 unchanged, 11
earlier and 11 later records without an identifier or a DOI not compared
# A tibble: 127 × 7
   status scopus_id doi                  title  year citations_old citations_new
   <fct>  <chr>     <chr>                <chr> <int>         <int>         <int>
 1 added  <NA>      10.1002/adfm.202315… Vasc…  2024            NA            32
 2 added  <NA>      10.1002/asia.202400… A Hi…  2024            NA             6
 3 added  <NA>      10.1002/slct.202302… Allo…  2024            NA             2
 4 added  <NA>      10.1016/j.cej.2024.… High…  2024            NA            31
 5 added  <NA>      10.1016/j.diamond.2… Copp…  2024            NA            21
 6 added  <NA>      10.1016/j.isci.2024… Boos…  2024            NA             8
 7 added  <NA>      10.1016/j.jallcom.2… Enha…  2024            NA             5
 8 added  <NA>      10.1016/j.jallcom.2… From…  2024            NA            84
 9 added  <NA>      10.1016/j.jpowsour.… How …  2024            NA            14
10 added  <NA>      10.1016/j.jpowsour.… High…  2024            NA             3
# ℹ 117 more rows

The citation columns show how much each record’s count changed between the two pulls. They are equal here, because both pulls were cut from one snapshot.

Merging without duplicates

To keep a cumulative set across retrievals, combine them. scopus_combine() renumbers the records and, with dedupe = TRUE, keeps each one once. Records with the same ‘Scopus’ identifier are one record, and a record without an identifier joins the identified record that carries its DOI. DOIs are compared ignoring case and any resolver prefix.

combined <- scopus_combine(baseline, later, dedupe = TRUE)
nrow(combined)
[1] 149

That is 149 rows for 138 distinct articles. These records did not come from ‘Scopus’ and carry no ‘Scopus’ identifier, so de-duplication falls back to the DOI. The eleven that arrived without one have no key to match on, so both copies of each are kept.

A live retrieval carries an identifier on every record, so the same call on two of them returns each record once. Records with different identifiers that share a DOI are both kept, since ‘Scopus’ gives one DOI to distinct papers as well as holding duplicate records of one paper. The merge counts the DOIs that such records share, and the search record gives this count so that the records can be checked by hand. Once they have been checked, dedupe = "id_or_doi" merges them.

c() on record sets calls scopus_combine(), so it concatenates and renumbers them in the same way and, unless it is given dedupe, keeps every row.

stacked <- c(baseline, later)
nrow(stacked)
[1] 261

Keeping a record of each pull

Saving each retrieval lets you compare against it next time. The .rds form round-trips exactly.

path <- file.path(tempdir(), "baseline.rds")
write_scopus_records(baseline, path)
identical(read_scopus_records(path), baseline)
[1] TRUE

A live retrieval also carries the date it was taken, as the retrieved_at attribute, together with the scopusflow version used to retrieve it. The date matters for this workflow in particular, because citations is a snapshot value that keeps changing, so a difference between two pulls means something only once you know how far apart they were. Both attributes survive the .rds form, and neither survives .csv. The bundled corpus used here carries neither.

In a live setting, the later retrieval would come from the API, and everything else would be as above. The baseline would be an earlier pull of your own, since the bundled one came from OpenAlex and would differ from any ‘Scopus’ pull for that reason alone. Both pulls would then be dated, and each would carry a ‘Scopus’ identifier on every record, so scopus_diff_records() could compare them all.

later <- scopus_fetch('"graphene supercapacitor"', field = "TITLE-ABS-KEY")
attr(later, "retrieved_at")
scopus_diff_dois(old = read_scopus_records(path), new = later)
scopus_diff_records(old = read_scopus_records(path), new = later)