Worldcloud figure in R

The following example presents a simple function to evaluate a chunk of text and present it in a fancy way: displaying a wordcloud.

The selected text is taken from the Bertrand Russell’s Nobel Prize Acceptance Speech

What Desires Are Politically Important?, 11th of December of 1950

The libraries used in this example are: wordcloud¹, tm², colortools³

    library(wordcloud)  # to analyse and plot text
    library(tm)         # a text mining package
    library(colortools) # to make colour functions

After loading the file, the function Corpus is applied to convert the plain text file to a natural language one, adding a metadata related to the existing words.

    speech = readLines("myData/BR.txt")
    myCorpus = Corpus(VectorSource(speech))

Then, a minimal mining process is performed in order to remove punctuation, numbers, symbols and common connectors.

    myCorpus <- tm_map(myCorpus, removePunctuation)
    myCorpus <- tm_map(myCorpus, removeNumbers)
    myCorpus <- tm_map(myCorpus, function(x)removeWords(x,stopwords("english")))

And finally, the text is summarised as a table of frequencies.

    tdm <- TermDocumentMatrix(myCorpus)
    m <- as.matrix(tdm)
    v <- sort(rowSums(m),decreasing=TRUE)
    d <- data.frame(word = names(v),freq=v)

Employing such a dataset, a graph can be generated using default parameters.

    wordcloud(d$word,d$freq,c(8,.3),2)

And as an illustrative example, the graph can be customized with extra parameters (..not the best plot though)

    wordcloud(d$word, d$freq, c(4, 0.5), 1, 300, TRUE, FALSE, .15, pal, vfont=c("gothic english","plain"))

NOTE: May I suggest an exercise? Try the function with a random ‘holy’ text such as xxxxxxx, words don’t lie. You may find too much controversial words for a bunch of sacred texts.

I hope this has been useful.

[- Download Speech]