In COMHIS/estc: English Short Title Catalogue (ESTC) Metadata Toolkit

Princeton summaries

# Load libraries
library(ggplot2)
library(tidyr)
library(dplyr)
library(reshape)
library(estc)
library(bibliographica)

# Complete data
df <- df.preprocessed
selected.topics <- c("Thirty Years' War, 1618-1648", "Popish Plot, 1678")
dfs <- filter(df, topic %in% selected.topics)

dfs <- dfs %>% group_by(topic, publication_decade) %>% summarize(paper = sum(paper, na.rm = TRUE), documents = n())

theme_set(theme_bw(20))
p <- ggplot(dfs, aes(x = publication_decade, y = documents, group = topic))
p <- p + geom_point(aes(col = topic))
p <- p + geom_line(aes(col = topic))
p <- p + xlab("Publication Decade")
p <- p + ylab("Documents (n)")
p1 <- p

p <- ggplot(dfs, aes(x = publication_decade, y = paper, group = topic))
p <- p + geom_point(aes(col = topic))
p <- p + geom_line(aes(col = topic))
p <- p + xlab("Publication Decade")
p <- p + ylab("Paper")
p2 <- p

library(gridExtra)
grid.arrange(p1, p2, nrow = 2)

#selected.countries <- c("England", "Ireland", "Scotland", "Wales")
selected.countries <- c("France", "Netherlands", "Germany", "Swizerland", "Belgium")
dfs <- filter(df, country %in% selected.countries)
dfs <- dfs %>% filter(!is.na(country))
dfs$country <- droplevels(dfs$country)

dfs <- dfs %>% group_by(country, publication_decade) %>% summarize(paper = sum(paper, na.rm = TRUE), documents = n())
theme_set(theme_bw(20))

p <- ggplot(dfs, aes(x = publication_decade, y = documents, group = country))
p <- p + geom_point(aes(col = country))
p <- p + geom_line(aes(col = country))
p <- p + xlab("Publication Decade")
p <- p + ylab("Documents (n)")
p1 <- p

p <- ggplot(dfs, aes(x = publication_decade, y = paper, group = country))
p <- p + geom_point(aes(col = country))
p <- p + geom_line(aes(col = country))
p <- p + xlab("Publication Decade")
p <- p + ylab("Paper")
p2 <- p

library(gridExtra)
grid.arrange(p1, p2, nrow = 2)