R: tidytext: sentiment comparing the three sentiment dictionaries

# Ref: https://github.com/dgrtwo/tidy-text-mining/blob/master/02-sentiment-analysis.Rmd

library(knitr)
opts_chunk$set(message = FALSE, warning = FALSE, cache = TRUE)
options(width = 100, dplyr.width = 100)
library(ggplot2)
theme_set(theme_light())

library(tidytext)
get_sentiments("afinn")
get_sentiments("bing")
get_sentiments("nrc")

pride_prejudice <- tidy_books %>% 
  filter(book == "Pride & Prejudice")
pride_prejudice

# use `inner_join()` to calculate the sentiment in different ways. 
afinn <- pride_prejudice %>% 
  inner_join(get_sentiments("afinn")) %>% 
  group_by(index = linenumber %/% 80) %>% 
  summarise(sentiment = sum(value)) %>% 
  mutate(method = "AFINN")
bing_and_nrc <- bind_rows(pride_prejudice %>% 
                            inner_join(get_sentiments("bing")) %>%
                            mutate(method = "Bing et al."),
                          pride_prejudice %>% 
                            inner_join(get_sentiments("nrc") %>% 
                                         filter(sentiment %in% c("positive", 
                                                                 "negative"))) %>%
                            mutate(method = "NRC")) %>%
  count(method, index = linenumber %/% 80, sentiment) %>%
  spread(sentiment, n, fill = 0) %>%
  mutate(sentiment = positive - negative)

afinn <- pride_prejudice %>% 
  inner_join(afinn) %>% 
  group_by(index = linenumber %/% 80) %>% 
  summarise(sentiment = sum(value)) %>% 
  mutate(method = "AFINN")
bing_and_nrc <- bind_rows(pride_prejudice %>% 
                            inner_join(get_sentiments("bing")) %>%
                            mutate(method = "Bing et al."),
                          pride_prejudice %>% 
                            inner_join(get_sentiments("nrc") %>% 
                                         filter(sentiment %in% c("positive", 
                                                                 "negative"))) %>%
                            mutate(method = "NRC")) %>%
  count(method, index = linenumber %/% 80, sentiment) %>%
  spread(sentiment, n, fill = 0) %>%
  mutate(sentiment = positive - negative)

# plot
bind_rows(afinn, 
          bing_and_nrc) %>%
  ggplot(aes(index, sentiment, fill = method)) +
  geom_col(show.legend = FALSE) +
  facet_wrap(~method, ncol = 1, scales = "free_y")

# count
get_sentiments("nrc") %>% 
  filter(sentiment %in% c("positive", 
                          "negative")) %>% 
  count(sentiment)
get_sentiments("bing") %>% 
  count(sentiment)

nrc %>% 
  filter(sentiment %in% c("positive", 
                          "negative")) %>% 
  count(sentiment)
get_sentiments("bing") %>% 
  count(sentiment)

Referensi

• https://github.com/dgrtwo/tidy-text-mining/blob/master/02-sentiment-analysis.Rmd

Pranala Menarik

• R