This is a working code in progress under the research project on the National Assembly of Republic of Korea, supported by the National Assembly Future Institute (NAFI). For more information, please contact inhwanko at uw dot edu.
library(tidyverse)
library(tidytext)
library(readxl)
library(quanteda)
library(readtext)
library(ggplot2)
library(rvest)bill_13 <- read_excel("bill_13.xlsx")
legislator_13 <- read_excel("legislator_13.xlsx")
legislator_13$party[legislator_13$party=="평화민주당"] <- "party1"
legislator_13$party[legislator_13$party=="통일민주당"] <- "party2"
legislator_13$party[legislator_13$party=="민주자유당"] <- "party3"
legislator_13$party[legislator_13$party=="민주정의당"] <- "party4"
legislator_13$party[legislator_13$party=="신민주공화당"] <- "party5"
legislator_13$party[legislator_13$party=="한겨레민주당평화민주당"] <- "party6"
legislator_13$party[legislator_13$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_13$submitter_text <- strsplit(bill_13$submitter_text, " ")
bill_13$supporter_text <- strsplit(bill_13$supporter_text, " ")
bill_13$legislator <- strsplit(bill_13$legislator, " ")
bill_13$noparty <- bill_13$party6 <- bill_13$party5 <- bill_13$party4 <- bill_13$party3 <- bill_13$party2 <- bill_13$party1 <- 0
bill_13$mainparty <- NA
bill_13 <- as.data.frame(bill_13)
for (i in 1:70) {
bill_13_legislator <- as.data.frame(unlist(bill_13$legislator[i]))
colnames(bill_13_legislator) <- "name"
bill_13_main <- as.data.frame(unlist(bill_13$submitter_text[i]))
colnames(bill_13_main) <- "name"
leftjoined <- left_join(bill_13_legislator, legislator_13[,3:4], by="name")
leftjoined_main <- left_join(bill_13_main, legislator_13[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
bill_13[i,]$mainparty <- match_main[1,]$party
try(bill_13[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_13[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_13[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_13[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_13[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_13[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_13[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_13[,c(-4,-5,-6)], "bill_13_final.csv")bill_14 <- read_excel("bill_14.xlsx")
legislator_14 <- read_excel("legislator_14.xlsx")
legislator_14$party[legislator_14$party=="민주당"] <- "party1"
legislator_14$party[legislator_14$party=="민주자유당"] <- "party2"
legislator_14$party[legislator_14$party=="신민당"] <- "party3"
legislator_14$party[legislator_14$party=="신정치개혁당"] <- "party4"
legislator_14$party[legislator_14$party=="통일민국당"] <- "party5"
legislator_14$party[legislator_14$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_14$submitter_text <- strsplit(bill_14$submitter_text, " ")
bill_14$supporter_text <- strsplit(bill_14$supporter_text, " ")
bill_14$legislator <- strsplit(bill_14$legislator, " ")
bill_14$noparty <- bill_14$party5 <- bill_14$party4 <- bill_14$party3 <- bill_14$party2 <- bill_14$party1 <- 0
bill_14$mainparty <- NA
bill_14 <- as.data.frame(bill_14)
for (i in 1:8) {
bill_14_legislator <- as.data.frame(unlist(bill_14$legislator[i]))
colnames(bill_14_legislator) <- "name"
bill_14_main <- as.data.frame(unlist(bill_14$submitter_text[i]))
colnames(bill_14_main) <- "name"
leftjoined <- left_join(bill_14_legislator, legislator_14[,3:4], by="name")
leftjoined_main <- left_join(bill_14_main, legislator_14[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_14[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_14[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_14[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_14[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_14[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_14[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_14[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_14[,c(-4,-5,-6)], "bill_14_final.csv")bill_15 <- read_excel("bill_15.xlsx")
legislator_15 <- read_excel("legislator_15.xlsx")
legislator_15$party[legislator_15$party=="새정치국민회의"] <- "party1"
legislator_15$party[legislator_15$party=="신한국당"] <- "party2"
legislator_15$party[legislator_15$party=="자유민주연합"] <- "party3"
legislator_15$party[legislator_15$party=="통합민주당"] <- "party4"
legislator_15$party[legislator_15$party=="한나라당"] <- "party5"
legislator_15$party[legislator_15$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_15$submitter_text <- strsplit(bill_15$submitter_text, " ")
bill_15$supporter_text <- strsplit(bill_15$supporter_text, " ")
bill_15$legislator <- strsplit(bill_15$legislator, " ")
bill_15$noparty <- bill_15$party5 <- bill_15$party4 <- bill_15$party3 <- bill_15$party2 <- bill_15$party1 <- 0
bill_15$mainparty <- NA
bill_15 <- as.data.frame(bill_15)
for (i in 1:59) {
bill_15_legislator <- as.data.frame(unlist(bill_15$legislator[i]))
colnames(bill_15_legislator) <- "name"
bill_15_main <- as.data.frame(unlist(bill_15$submitter_text[i]))
colnames(bill_15_main) <- "name"
leftjoined <- left_join(bill_15_legislator, legislator_15[,3:4], by="name")
leftjoined_main <- left_join(bill_15_main, legislator_15[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_15[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_15[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_15[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_15[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_15[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_15[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_15[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_15[,c(-4,-5,-6)], "bill_15_final.csv")bill_16 <- read_excel("bill_16.xlsx")
legislator_16 <- read_excel("legislator_16.xlsx")
legislator_16$party[legislator_16$party=="개혁국민정당"] <- "party1"
legislator_16$party[legislator_16$party=="민주국민당"] <- "party2"
legislator_16$party[legislator_16$party=="새천년민주당"] <- "party3"
legislator_16$party[legislator_16$party=="자유민주연합"] <- "party4"
legislator_16$party[legislator_16$party=="한나라당"] <- "party5"
legislator_16$party[legislator_16$party=="희망의 한국신당"] <- "party6"
legislator_16$party[legislator_16$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_16$submitter_text <- strsplit(bill_16$submitter_text, " ")
bill_16$supporter_text <- strsplit(bill_16$supporter_text, " ")
bill_16$legislator <- strsplit(bill_16$legislator, " ")
bill_16$noparty <- bill_16$party6 <- bill_16$party5 <- bill_16$party4 <- bill_16$party3 <- bill_16$party2 <- bill_16$party1 <- 0
bill_16$mainparty <- NA
bill_16 <- as.data.frame(bill_16)
for (i in 1:214) {
bill_16_legislator <- as.data.frame(unlist(bill_16$legislator[i]))
colnames(bill_16_legislator) <- "name"
bill_16_main <- as.data.frame(unlist(bill_16$submitter_text[i]))
colnames(bill_16_main) <- "name"
leftjoined <- left_join(bill_16_legislator, legislator_16[,3:4], by="name")
leftjoined_main <- left_join(bill_16_main, legislator_16[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_16[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_16[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_16[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_16[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_16[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_16[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_16[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_16[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_16[,c(-4,-5,-6)], "bill_16_final.csv")bill_17 <- read_excel("bill_17.xlsx")
legislator_17 <- read_excel("legislator_17.xlsx")
legislator_17$party[legislator_17$party=="국민중심당"] <- "party1"
legislator_17$party[legislator_17$party=="국민통합21"] <- "party2"
legislator_17$party[legislator_17$party=="대통합민주신당"] <- "party3"
legislator_17$party[legislator_17$party=="민주노동당"] <- "party4"
legislator_17$party[legislator_17$party=="민주당"] <- "party5"
legislator_17$party[legislator_17$party=="새천년민주당"] <- "party6"
legislator_17$party[legislator_17$party=="열린우리당"] <- "party7"
legislator_17$party[legislator_17$party=="자유민주연합"] <- "party8"
legislator_17$party[legislator_17$party=="통합민주당"] <- "party9"
legislator_17$party[legislator_17$party=="한나라당"] <- "party10"
legislator_17$party[legislator_17$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_17$submitter_text <- strsplit(bill_17$submitter_text, " ")
bill_17$supporter_text <- strsplit(bill_17$supporter_text, " ")
bill_17$legislator <- strsplit(bill_17$legislator, " ")
bill_17$noparty <- bill_17$party10 <- bill_17$party9 <- bill_17$party8 <- bill_17$party7 <- bill_17$party6 <- bill_17$party5 <- bill_17$party4 <- bill_17$party3 <- bill_17$party2 <- bill_17$party1 <- 0
bill_17$mainparty <- NA
bill_17 <- as.data.frame(bill_17)
for (i in 1:735) {
bill_17_legislator <- as.data.frame(unlist(bill_17$legislator[i]))
colnames(bill_17_legislator) <- "name"
bill_17_main <- as.data.frame(unlist(bill_17$submitter_text[i]))
colnames(bill_17_main) <- "name"
leftjoined <- left_join(bill_17_legislator, legislator_17[,3:4], by="name")
leftjoined_main <- left_join(bill_17_main, legislator_17[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_17[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_17[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_17[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_17[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_17[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_17[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_17[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_17[i,]$party7 <- match$n[match$party=="party7"], silent=T)
try(bill_17[i,]$party8 <- match$n[match$party=="party8"], silent=T)
try(bill_17[i,]$party9 <- match$n[match$party=="party9"], silent=T)
try(bill_17[i,]$party10 <- match$n[match$party=="party10"], silent=T)
try(bill_17[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_17[,c(-4,-5,-6)], "bill_17_final.csv")bill_18 <- read_excel("bill_18.xlsx")
legislator_18 <- read_excel("legislator_18.xlsx")
legislator_18$party[legislator_18$party=="민주노동당"] <- "party1"
legislator_18$party[legislator_18$party=="민주당"] <- "party2"
legislator_18$party[legislator_18$party=="민주통합당"] <- "party3"
legislator_18$party[legislator_18$party=="새누리당"] <- "party4"
legislator_18$party[legislator_18$party=="자유선진당"] <- "party5"
legislator_18$party[legislator_18$party=="진보신당"] <- "party6"
legislator_18$party[legislator_18$party=="창조한국당"] <- "party7"
legislator_18$party[legislator_18$party=="친박연대"] <- "party8"
legislator_18$party[legislator_18$party=="통합민주당"] <- "party9"
legislator_18$party[legislator_18$party=="통합진보당"] <- "party10"
legislator_18$party[legislator_18$party=="한나라당"] <- "party11"
legislator_18$party[legislator_18$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_18$submitter_text <- strsplit(bill_18$submitter_text, " ")
bill_18$supporter_text <- strsplit(bill_18$supporter_text, " ")
bill_18$legislator <- strsplit(bill_18$legislator, " ")
bill_18$noparty <- bill_18$party11 <- bill_18$party10 <- bill_18$party9 <- bill_18$party8 <- bill_18$party7 <- bill_18$party6 <- bill_18$party5 <- bill_18$party4 <- bill_18$party3 <- bill_18$party2 <- bill_18$party1 <- 0
bill_18$mainparty <- NA
bill_18 <- as.data.frame(bill_18)
for (i in 1:1098) {
bill_18_legislator <- as.data.frame(unlist(bill_18$legislator[i]))
colnames(bill_18_legislator) <- "name"
bill_18_main <- as.data.frame(unlist(bill_18$submitter_text[i]))
colnames(bill_18_main) <- "name"
leftjoined <- left_join(bill_18_legislator, legislator_18[,3:4], by="name")
leftjoined_main <- left_join(bill_18_main, legislator_18[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_18[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_18[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_18[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_18[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_18[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_18[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_18[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_18[i,]$party7 <- match$n[match$party=="party7"], silent=T)
try(bill_18[i,]$party8 <- match$n[match$party=="party8"], silent=T)
try(bill_18[i,]$party9 <- match$n[match$party=="party9"], silent=T)
try(bill_18[i,]$party10 <- match$n[match$party=="party10"], silent=T)
try(bill_18[i,]$party11 <- match$n[match$party=="party11"], silent=T)
try(bill_18[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_18[,c(-4,-5,-6)], "bill_18_final.csv")bill_19 <- read_excel("bill_19.xlsx")
legislator_19 <- read_excel("legislator_19.xlsx")
legislator_19$party[legislator_19$party=="더불어민주당"] <- "party1"
legislator_19$party[legislator_19$party=="민주당"] <- "party2"
legislator_19$party[legislator_19$party=="민주통합당"] <- "party3"
legislator_19$party[legislator_19$party=="새누리당"] <- "party4"
legislator_19$party[legislator_19$party=="새정치민주연합"] <- "party5"
legislator_19$party[legislator_19$party=="자유선진당"] <- "party6"
legislator_19$party[legislator_19$party=="통합진보당"] <- "party7"
legislator_19$party[legislator_19$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_19$submitter_text <- strsplit(bill_19$submitter_text, " ")
bill_19$supporter_text <- strsplit(bill_19$supporter_text, " ")
bill_19$legislator <- strsplit(bill_19$legislator, " ")
bill_19$noparty <- bill_19$party7 <- bill_19$party6 <- bill_19$party5 <- bill_19$party4 <- bill_19$party3 <- bill_19$party2 <- bill_19$party1 <- 0
bill_19$mainparty <- NA
bill_19 <- as.data.frame(bill_19)
for (i in 1:1488) {
bill_19_legislator <- as.data.frame(unlist(bill_19$legislator[i]))
colnames(bill_19_legislator) <- "name"
bill_19_main <- as.data.frame(unlist(bill_19$submitter_text[i]))
colnames(bill_19_main) <- "name"
leftjoined <- left_join(bill_19_legislator, legislator_19[,3:4], by="name")
leftjoined_main <- left_join(bill_19_main, legislator_19[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_19[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_19[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_19[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_19[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_19[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_19[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_19[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_19[i,]$party7 <- match$n[match$party=="party7"], silent=T)
try(bill_19[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_19[,c(-4,-5,-6)], "bill_19_final.csv")bill_20 <- read_excel("bill_20.xlsx")
legislator_20 <- read_excel("legislator_20.xlsx")
legislator_20$party[legislator_20$party=="국민의당"] <- "party1"
legislator_20$party[legislator_20$party=="더불어민주당"] <- "party2"
legislator_20$party[legislator_20$party=="미래통합당"] <- "party3"
legislator_20$party[legislator_20$party=="민생당"] <- "party4"
legislator_20$party[legislator_20$party=="바른미래당"] <- "party5"
legislator_20$party[legislator_20$party=="새누리당"] <- "party6"
legislator_20$party[legislator_20$party=="자유한국당"] <- "party7"
legislator_20$party[legislator_20$party=="정의당"] <- "party8"
legislator_20$party[legislator_20$party=="무소속"] <- "noparty"
# Transform each submitter and supporter into string vectors
bill_20$submitter_text <- strsplit(bill_20$submitter_text, " ")
bill_20$supporter_text <- strsplit(bill_20$supporter_text, " ")
bill_20$legislator <- strsplit(bill_20$legislator, " ")
bill_20$noparty <- bill_20$party8 <-bill_20$party7 <- bill_20$party6 <- bill_20$party5 <- bill_20$party4 <- bill_20$party3 <- bill_20$party2 <- bill_20$party1 <- 0
bill_20$mainparty <- NA
bill_20 <- as.data.frame(bill_20)
for (i in 1:1954) {
bill_20_legislator <- as.data.frame(unlist(bill_20$legislator[i]))
colnames(bill_20_legislator) <- "name"
bill_20_main <- as.data.frame(unlist(bill_20$submitter_text[i]))
colnames(bill_20_main) <- "name"
leftjoined <- left_join(bill_20_legislator, legislator_20[,3:4], by="name")
leftjoined_main <- left_join(bill_20_main, legislator_20[,3:4], by="name")
match <- leftjoined %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
match_main <- leftjoined_main %>%
na.omit() %>%
group_by(party) %>%
summarize(n=n()) %>%
arrange(desc(n))
try(bill_20[i,]$mainparty <- match_main[1,]$party, silent=T)
try(bill_20[i,]$party1 <- match$n[match$party=="party1"], silent=T)
try(bill_20[i,]$party2 <- match$n[match$party=="party2"], silent=T)
try(bill_20[i,]$party3 <- match$n[match$party=="party3"], silent=T)
try(bill_20[i,]$party4 <- match$n[match$party=="party4"], silent=T)
try(bill_20[i,]$party5 <- match$n[match$party=="party5"], silent=T)
try(bill_20[i,]$party6 <- match$n[match$party=="party6"], silent=T)
try(bill_20[i,]$party7 <- match$n[match$party=="party7"], silent=T)
try(bill_20[i,]$party8 <- match$n[match$party=="party8"], silent=T)
try(bill_20[i,]$noparty <- match$n[match$party=="noparty"], silent=T)
}
write.csv(bill_20[,c(-4,-5,-6)], "bill_20_final.csv")