Bible Text Mining in R Part 2

Load libraries

library(tm)
library(wordcloud)
library(RColorBrewer)
library(RSentiment)
library(rJava)

Read corpus from part 1

docs <- readRDS(file = "docs.rds")

Create document term matrix and get frequency

bible_dtm <-DocumentTermMatrix(VCorpus(VectorSource(docs[[1]]$content)))
freq <- colSums(as.matrix(bible_dtm))

Download JAVA and fix issue with tokeniizng

Sys.setenv(JAVA_HOME='C:\\Program Files\\Java\\jre1.8.0_231')

Calculating frequency of sentiment and combining it to the dataframe

sentiments_bible <- calculate_sentiment(names(freq))
sentiments_bible <- cbind(sentiments_bible, as.data.frame(freq))

Categorising words into negative, neutral, and positive

sent_pos_bible <- sentiments_bible[sentiments_bible$sentiment == 'Positive',]
sent_neu_bible <- sentiments_bible[sentiments_bible$sentiment == 'Neutral',]
sent_neg_bible <- sentiments_bible[sentiments_bible$sentiment == 'Negative',]

Plot top positive words

set.seed(100)
wordcloud(sent_pos_bible$text,sent_pos_bible$freq, min.freq=400,colors=brewer.pal(6,"Dark2"), random.order = F, random.color = F)

Plot top neutral words

set.seed(100)
wordcloud(sent_neu_bible$text,sent_neu_bible$freq, min.freq=1000,colors=brewer.pal(6,"Dark2"), random.order = F, random.color = F)

Plot top negative words

set.seed(100)
wordcloud(sent_neg_bible$text,sent_neg_bible$freq, min.freq=100000000**1000000000000,colors=brewer.pal(20,"Dark2"), random.order = F, random.color = F)