Introduction to tidytext
In tidytext: Text Mining using 'dplyr', 'ggplot2', and Other Tidy Tools

#| label = "setup",
#| echo = FALSE
library(knitr)
opts_chunk$set(
  warning = FALSE, message = FALSE,
  eval = requireNamespace("wordcloud", quietly = TRUE) && requireNamespace("ggplot2", quietly = TRUE)
  )

#| echo = FALSE
library(ggplot2)
theme_set(theme_light())

The Life-Changing Magic of Tidying Text

Using tidy data principles can make many text mining tasks easier, more effective, and consistent with tools already in wide use. Much of the infrastructure needed for text mining with tidy data frames already exists in packages like dplyr, broom, tidyr and ggplot2. In this package, we provide functions and supporting data sets to allow conversion of text to and from tidy formats, and to switch seamlessly between tidy tools and existing text mining packages. Check out our book to learn more about text mining using tidy data principles.

A few first tidy text mining examples

The novels of Jane Austen can be so tidy! Let's use the text of Jane Austen's 6 completed, published novels from the janeaustenr package, and transform them into a tidy format. janeaustenr provides them as a one-row-per-line format:

library(janeaustenr)
library(dplyr)
library(stringr)

original_books <- austen_books() %>%
  group_by(book) %>%
  mutate(line = row_number(),
         chapter = cumsum(str_detect(text, regex("^chapter [\\divxlc]",
                                                 ignore_case = TRUE)))) %>%
  ungroup()

original_books

To work with this as a tidy dataset, we need to restructure it as one-token-per-row format. The unnest_tokens function is a way to convert a dataframe with a text column to be one-token-per-row. Here let's tokenize to a new word column from the existing text column:

library(tidytext)
tidy_books <- original_books %>%
  unnest_tokens(output = word, input = text)

tidy_books

This function uses the tokenizers package to separate each line into words. The default tokenizing is for words, but other options include characters, ngrams, sentences, lines, paragraphs, or separation around a regex pattern.

Now that the data is in one-word-per-row format, we can manipulate it with tidy tools like dplyr. We can remove stop words (accessible in a tidy form with the function get_stopwords()) with an anti_join.

cleaned_books <- tidy_books %>%
  anti_join(get_stopwords())

We can also use count to find the most common words in all the books as a whole.

cleaned_books %>%
  count(word, sort = TRUE)

Sentiment analysis can be done as an inner join. Sentiment lexicons are available via the get_sentiments() function. Let's look at the words with a positive score from the lexicon of Bing Liu and collaborators. What are the most common positive words in Emma?

positive <- get_sentiments("bing") %>%
  filter(sentiment == "positive")

tidy_books %>%
  filter(book == "Emma") %>%
  semi_join(positive) %>%
  count(word, sort = TRUE)

Or instead we could examine how sentiment changes during each novel. Let's find a sentiment score for each word using the same lexicon, then count the number of positive and negative words in defined sections of each novel.

library(tidyr)
bing <- get_sentiments("bing")

janeaustensentiment <- tidy_books %>%
  inner_join(bing, relationship = "many-to-many") %>%
  count(book, index = line %/% 80, sentiment) %>%
  pivot_wider(names_from = sentiment, values_from = n, values_fill = 0) %>% 
  mutate(sentiment = positive - negative)

Now we can plot these sentiment scores across the plot trajectory of each novel.

#| fig.width = 7,
#| fig.height = 7,
#| fig.alt = "Sentiment scores across the trajectories of Jane Austen's six published novels",
#| warning = FALSE
library(ggplot2)

ggplot(janeaustensentiment, aes(index, sentiment, fill = book)) +
  geom_bar(stat = "identity", show.legend = FALSE) +
  facet_wrap(vars(book), ncol = 2, scales = "free_x")

Most common positive and negative words

One advantage of having the data frame with both sentiment and word is that we can analyze word counts that contribute to each sentiment.

bing_word_counts <- tidy_books %>%
  inner_join(bing, relationship = "many-to-many") %>%
  count(word, sentiment, sort = TRUE)

bing_word_counts

This can be shown visually, and we can pipe straight into ggplot2 because of the way we are consistently using tools built for handling tidy data frames.

#| fig.width = 7,
#| fig.height = 4,
#| fig.alt = 'Bar charts for the contribution of words to sentiment scores. The words "well" and "good" contribute the most to positive sentiment, and the word "miss" contributes the most to negative sentiment'
bing_word_counts %>%
  group_by(sentiment) %>%
  slice_max(n, n = 10) %>%
  mutate(word = reorder(word, n)) %>%
  ggplot(aes(n, word, fill = sentiment)) +
  geom_col(show.legend = FALSE) +
  facet_wrap(vars(sentiment), scales = "free_y") +
  labs(x = "Contribution to sentiment", y = NULL)

This lets us spot an anomaly in the sentiment analysis; the word "miss" is coded as negative but it is used as a title for young, unmarried women in Jane Austen's works. If it were appropriate for our purposes, we could easily add "miss" to a custom stop-words list using bind_rows.

Wordclouds

We've seen that this tidy text mining approach works well with ggplot2, but having our data in a tidy format is useful for other plots as well.

For example, consider the wordcloud package. Let's look at the most common words in Jane Austen's works as a whole again.

#| fig.height = 6,
#| fig.width = 6,
#| fig.alt = 'Wordcloud showing that words like "mr", "miss", and "one" are most common in the works of Jane Austen'
library(wordcloud)

cleaned_books %>%
  count(word) %>%
  with(wordcloud(word, n, max.words = 100))

In other functions, such as comparison.cloud, you may need to turn it into a matrix with reshape2's acast. Let's do the sentiment analysis to tag positive and negative words using an inner join, then find the most common positive and negative words. Until the step where we need to send the data to comparison.cloud, this can all be done with joins, piping, and dplyr because our data is in tidy format.

#| label = "wordcloud",
#| fig.height = 5,
#| fig.width = 5,
#| fig.alt = 'Wordcloud showing that "well" and "good" are the most common positive sentiment words while "miss" is the most common negative sentiment word'
library(reshape2)

tidy_books %>%
  inner_join(bing) %>%
  count(word, sentiment, sort = TRUE) %>%
  acast(word ~ sentiment, value.var = "n", fill = 0) %>%
  comparison.cloud(colors = c("#F8766D", "#00BFC4"),
                   max.words = 100)

Looking at units beyond just words

Lots of useful work can be done by tokenizing at the word level, but sometimes it is useful or necessary to look at different units of text. For example, some sentiment analysis algorithms look beyond only unigrams (i.e. single words) to try to understand the sentiment of a sentence as a whole. These algorithms try to understand that

I am not having a good day.

is a sad sentence, not a happy one, because of negation. The Stanford CoreNLP tools and the sentimentr R package are examples of such sentiment analysis algorithms. For these, we may want to tokenize text into sentences.

PandP_sentences <- tibble(text = prideprejudice) %>% 
  unnest_tokens(output = sentence, input = text, token = "sentences")

Let's look at just one.

PandP_sentences$sentence[2]

The sentence tokenizing does seem to have a bit of trouble with UTF-8 encoded text, especially with sections of dialogue; it does much better with punctuation in ASCII.

Another option in unnest_tokens is to split into tokens using a regex pattern. We could use this, for example, to split the text of Jane Austen's novels into a data frame by chapter.

austen_chapters <- austen_books() %>%
  group_by(book) %>%
  unnest_tokens(chapter, text, token = "regex", pattern = "Chapter|CHAPTER [\\dIVXLC]") %>%
  ungroup()

austen_chapters %>% 
  group_by(book) %>% 
  summarise(chapters = n())

We have recovered the correct number of chapters in each novel (plus an "extra" row for each novel title). In this data frame, each row corresponds to one chapter.

Near the beginning of this vignette, we used a similar regex to find where all the chapters were in Austen's novels for a tidy data frame organized by one-word-per-row. We can use tidy text analysis to ask questions such as what are the most negative chapters in each of Jane Austen's novels? First, let's get the list of negative words from the Bing lexicon. Second, let's make a dataframe of how many words are in each chapter so we can normalize for the length of chapters. Then, let's find the number of negative words in each chapter and divide by the total words in each chapter. Which chapter has the highest proportion of negative words?

bingnegative <- get_sentiments("bing") %>%
  filter(sentiment == "negative")

wordcounts <- tidy_books %>%
  group_by(book, chapter) %>%
  summarize(words = n())

tidy_books %>%
  semi_join(bingnegative) %>%
  group_by(book, chapter) %>%
  summarize(negativewords = n()) %>%
  left_join(wordcounts, by = c("book", "chapter")) %>%
  mutate(ratio = negativewords/words) %>%
  filter(chapter != 0) %>%
  slice_max(ratio, n = 1)

These are the chapters with the most negative words in each book, normalized for number of words in the chapter. What is happening in these chapters? In Chapter 43 of Sense and Sensibility Marianne is seriously ill, near death, and in Chapter 34 of Pride and Prejudice Mr. Darcy proposes for the first time (so badly!). Chapter 46 of Mansfield Park is almost the end, when everyone learns of Henry's scandalous adultery, Chapter 15 of Emma is when horrifying Mr. Elton proposes, and in Chapter 21 of Northanger Abbey Catherine is deep in her Gothic faux fantasy of murder, etc. Chapter 4 of Persuasion is when the reader gets the full flashback of Anne refusing Captain Wentworth and how sad she was and what a terrible mistake she realized it to be.