preprint_citations.Rmd

---
title: "R Notebook"
---

```{r}

library(tidyverse)
library(rdimensions) # in development: https://github.com/nicholasmfraser/rdimensions

# Retrieve auth token for dimensions
dimensions_login()

```

# Load preprint data

```{r}

preprints <- read_csv("data/preprints_full_20190101_20201031.csv")

```

# Retrieve citations for preprints and their respective published articles

```{r}

# Function for retrieving citations for a set of DOIs from Dimensions
getCitations <- function(dois) {
  dois <- map_chr(dois, ~str_c('"', ., '"'))
  doi_string <- str_c(dois, collapse = ", ")
  q <- enc2utf8(str_c('search publications where doi in [', 
                      doi_string, 
                      '] return publications[doi+times_cited] limit 1000'))
  data <- dimensions_query(noquote(q))$publications
  Sys.sleep(1)
  return(tibble(
    doi = map_chr(data, ~.$doi),
    citations = map_int(data, ~.$times_cited)
  ))
}

max_dois <- 250 # Max DOIs allowable per query

# Citations for preprints
preprint_dois <- preprints %>% pull(doi)
preprint_doi_chunks <- split(preprint_dois, ceiling(seq_along(preprint_dois)/max_dois))
preprint_citations <- map_dfr(preprint_doi_chunks, getCitations) %>%
  mutate(doi = str_trim(str_to_lower(doi)))

# Citations for published_articles
published_dois <- preprints %>% filter(!is.na(published_doi)) %>% pull(published_doi)
published_doi_chunks <- split(published_dois, ceiling(seq_along(published_dois)/max_dois))
published_citations <- map_dfr(published_doi_chunks, getCitations) %>%
  rename(
    published_doi = doi,
    published_citations = citations
  ) %>%
  group_by(published_doi) %>% 
  # In a small number of cases, multiple records are returned for the same article. 
  # In each group, one article usually has citation counts and the others do not
  # here we select the one with the highest citation counts
  arrange(desc(published_citations)) %>% 
  slice(1) %>%
  mutate(published_doi = str_trim(str_to_lower(published_doi)))

```

# Create final citations dataset

```{r}

preprints %>%
  left_join(preprint_citations, by = "doi") %>%
  # when a preprint is not found, we assume it has zero citations
  replace_na(list(citations = 0)) %>%
  left_join(published_citations, by = "published_doi") %>%
  select(doi, citations, published_doi, published_citations) %>%
  write_csv("data/preprint_citations_20190101_20201031.csv")

```