emotion: Compute Emotion Rate

Description

Detect the rate of emotion at the sentence level. This method uses a simple dictionary lookup to find emotion words and then compute the rate per sentence. The emotion score ranges between 0 (no emotion used) and 1 (all words used were emotional). Note that a single emotion phrase would count as just one in the emotion_count column but would count as two words in the word_count column.

Usage

emotion(text.var, emotion_dt = lexicon::hash_nrc_emotions,
  valence_shifters_dt = lexicon::hash_valence_shifters,
  drop.unused.emotions = FALSE, un.as.negation = TRUE,
  un.as.negation.warn = isTRUE(all.equal(valence_shifters_dt,
  lexicon::hash_nrc_emotions)), n.before = 5, n.after = 2, ...)

Arguments

text.var

The text variable. Can be a get_sentences object or a raw character vector though get_sentences is preferred as it avoids the repeated cost of doing sentence boundary disambiguation every time sentiment is run.

emotion_dt

A data.table with a token and emotion column (tokens are nested within the emotions. The table cannot contain any duplicate rows and must have the token column set as the key column (see ?data.table::setkey). The default emotion table is lexicon::hash_nrc_emotions.

valence_shifters_dt

A data.table of valence shifters that can alter a polarized word's meaning and an integer key for negators (1), amplifiers [intensifiers] (2), de-amplifiers [downtoners] (3) and adversative conjunctions (4) with x and y as column names. For this purpose only negators is required/used.

drop.unused.emotions

logical. If TRUE unused/unfound emotion levels will not be included in the output.

un.as.negation

logical. If TRUE then emotion words prefixed with an 'un-' are treated as a negation. For example,"unhappy" would be treated as "not happy". If an emotion word has an un- version in the emotion_dt then no substitution is performed and an optional warning will be given.

un.as.negation.warn

logical. If TRUE and if un.as.negation id TRUE, then a warning will be given if the -un version of an emotion term is already found within the emotion_dt. Note that the default emotion_dt, lexicon::hash_nrc_emotions, will not give a warning unless it is explicitly set to do so. There are a number of emotion words in lexicon::hash_nrc_emotions that contain un- prefixed versions already in the dictionary. Use: emotion('', un.as.negation.warn = TRUE) to see these un- prefixed emotion words that are contained within lexicon::hash_nrc_emotions.

n.before

The number of words to consider as negated before the emotion word. To consider the entire beginning portion of a sentence use n.before = Inf. Note that a comma, colon, or semicolon acts as a boundary for considered words. Only words between the emotion word and these punctuation types will be considered.

n.after

The number of words to consider as negated after the emotion word. To consider the entire ending portion of a sentence use n.after = Inf. Note that a comma, colon, or semicolon acts as a boundary for considered words. Only words between the emotion word and these punctuation types will be considered.

…

ignored.

Value

Returns a data.table of:

element_id - The id number of the original vector passed to emotion
sentence_id - The id number of the sentences within each element_id
word_count - Word count
emotion_type - Type designation from the emotion column of the emotion_dt table
emotion_count - Count of the number of emotion words of that emotion_type
emotion - A score of the percentage of emotion words of that emotion_type

References

Plutchik, R. (1962). The emotions: Facts and theories, and a new model. Random House studies in psychology. Random House. Plutchik, R. (2001). The nature of emotions: Human emotions have deep evolutionary roots, a fact that may explain their complexity and provide tools for clinical practice. American Scientist , 89 (4), 344-350.

Examples

Run this code

# NOT RUN {
mytext <- c(
    "I am not afraid of you",
    NA,
    "",
    "I love it [not really]", 
    "I'm not angry with you", 
    "I hate it when you lie to me.  It's so humiliating",
    "I'm not happpy anymore.  It's time to end it",
    "She's a darn good friend to me",
    "I went to the terrible store",
    "There is hate and love in each of us",
    "I'm no longer angry!  I'm really experiencing peace but not true joy.",
    
    paste("Out of the night that covers me, Black as the Pit from pole to", 
      "pole, I thank whatever gods may be For my unconquerable soul."
     ),
    paste("In the fell clutch of circumstance I have not winced nor cried",
        "aloud. Under the bludgeonings of chance My head is bloody, but unbowed."
    ),
    paste("Beyond this place of wrath and tears Looms but the Horror of the", 
        "shade, And yet the menace of the years Finds, and shall find, me unafraid."
    ),
    paste("It matters not how strait the gate, How charged with punishments", 
        "the scroll, I am the master of my fate: I am the captain of my soul."
    )    
    
)

## works on a character vector but not the preferred method avoiding the 
## repeated cost of doing sentence boundary disambiguation every time 
## `emotion` is run
emotion(mytext)

## preferred method avoiding paying the cost 
split_text <- get_sentences(mytext)
(emo <- emotion(split_text))
emotion(split_text, drop.unused.emotions = TRUE)

# }
# NOT RUN {
plot(emo)
plot(emo, drop.unused.emotions = FALSE)
plot(emo, facet = FALSE)
plot(emo, facet = 'negated')

library(data.table)
fear <- emo[
    emotion_type == 'fear', ][, 
    text := unlist(split_text)][]
    
fear[emotion > 0,]

brady <- get_sentences(crowdflower_deflategate)
brady_emotion <- emotion(brady)
brady_emotion
# }