ニコニコ動画コメント等データから、動画コメントを取す例2 http://www.nii.ac.jp/cscenter/idr/nico/nico-apply.html

gistfile1.r

# 動画コメントを取り出す例 2

# 動画コメントのjsonデータは行単位になっているが、どうも扱いづらいので、jsonの配列に変換しておく。
# 
# 元
# {"date":1,"no":2,"vpos":3,"comment":"hello!","command":"184"}
# {"date":2,"no":3,"vpos":4,"comment":"hello!!","command":"184"}
# …
# 
# これを変換して
# [{"date":1,"no":2,"vpos":3,"comment":"hello!","command":"184"},
# {"date":2,"no":3,"vpos":4,"comment":"hello!!","command":"184"},
# …]
# 
# としておく。
# 
# もっとまともな方法が欲しい。

#install.packages("rjson")
library("rjson")
json_file <- "/ファイル/thread/sm15108950_json.dat"
json_data <- fromJSON(paste(readLines(json_file), collapse=""))

comment <- c()
comment_length <- c()
vpos <- c()
# for ( i in 1:length(json_data) ) {
for ( i in 1:10000 ) {
  comment <- c( comment, json_data[[i]]$comment[1] )
	comment_length <- c( comment_length, nchar(json_data[[i]]$comment[1] ) )
	vpos <- c( vpos, json_data[[i]]$vpos[1] )
}

# vposは1/100secなので、分にする
plot(
	head(vpos, n=10000)/100/60, head(comment_length, n=10000) ,
	main = "Alice in Musicland (sm15108950, 10000 comments)",
	xlab = "vpos (min)",
	ylab = "length(comment)",
	xlim=c(0,12.5), ylim=c(0,80)
)


#lowessを重ねる

#f=2/3
par(new=T)
lines(
	lowess( head(vpos, n=10000)/100/60, head(comment_length, n=10000), f=2/3),
	main = "Alice in Musicland (sm15108950, 10000 comments)",
	xlab = "vpos (min)",
	ylab = "length(comment)",
	xlim=c(0,12.5), ylim=c(0,80),
	col = "red"
)

#f=0.2
par(new=T)
lines(
	lowess( head(vpos, n=10000)/100/60, head(comment_length, n=10000), f=0.2),
	main = "Alice in Musicland (sm15108950, 10000 comments)",
	xlab = "vpos (min)",
	ylab = "length(comment)",
	xlim=c(0,12.5), ylim=c(0,80),
	col = "green"
)

#label
legend(0, 80, c(paste("f = ", c("2/3", "0.2"))), lty = 1, col = 2:3)