Unpaywall evidence type

unpaywall_evidence.R

#' bq unpaywall dump
#' 
#' connect to google bg with imported jsonl Unpaywall dump
library(DBI)
library(tidyverse)
library(bigrquery)
con <- dbConnect(
  bigrquery::bigquery(),
  project = "api-project-764811344545",
  dataset = "oadoi_full"
)
#' connection
oadoi <- tbl(con, "mongo_13_sep_18")
#' query (SQL)
my_query <- c("SELECT evidence, is_best, year, COUNT(*) AS number_of_articles
FROM `oadoi_full.mongo_13_sep_18`, UNNEST(oa_locations)
              GROUP BY evidence, year, is_best")
#' call bq
dbGetQuery(con, my_query) -> my_evidence
#' plot
my_evidence %>%
  filter(year %in% 2013:2018) %>%
  ggplot(aes(factor(year), number_of_articles, fill = is_best)) +
  geom_bar(stat = "identity") +
  facet_wrap(~evidence, ncol = 3) +
  scale_y_continuous(
    labels = function(x)
      format(x, big.mark = " ", scientific = FALSE),
    breaks = scales::pretty_breaks()
  ) +
  scale_fill_manual(
    values = c("#23445D","#AE4132")
  ) +
  theme_minimal() +
  labs(x = "Year", y = "Articles", title = "Unpaywall OA evidence categories per year")