Wordcloud em R

wordcloud-in-r.R

library(RColorBrewer)
library(wordcloud)

pal2 <- brewer.pal(8,"Set2")

lines <- readLines("pg32519.txt")
lines <- lines[45:3192]

text <- paste(lines, collapse = " ")
words <- unlist(strsplit(tolower(text), "\\s+|,|\\(|\\)|:|\\.|!|\\?|;"))

stopwords <- c("para", "from","que","não","por","mais","com","seus",
               "seu","uma","sua", "pelos","assim","estava","então",
               "contra","muitos","agora","entre",
               "tanto","tambem","havia","haviam","quase","sempre","colombo")

pos <- c()
for(w in stopwords)
  pos <- c(pos, which(words == w))

words <- words[-pos]

words <- words[nchar(words) > 3]
twords <- table(words)

# To check if I need more stopwords
# tmp <- as.data.frame(twords)
# head(tmp[order(tmp$Freq,decreasing=T),], 20)

# Removing the words that occurs less than 2 times
twords <- twords[twords > 2]

# the plot
png("cloud.png", width=580, height=580)
wordcloud(names(twords), twords, scale=c(9,.1),min.freq=2, max.words=Inf,
          random.order=F, rot.per=.3, colors=pal2)
dev.off()