tidyEmoji

Count, score, translate and relate the emoji in any text column, and get tidy data frames back.

library(tidyEmoji)

reviews <- data.frame(text = c("Best purchase ever \U0001f600\U0001f60d",
                               "It broke after a day \U0001f621",
                               "Does the job.",
                               "Shipped to \U0001f1ef\U0001f1f5 in two days \U0001f680"))

emoji_frequency(reviews, text)
#> # A tibble: 5 × 5
#>   emoji name                         shortcode  group                 n
#>   <chr> <chr>                        <chr>      <chr>             <int>
#> 1 🇯🇵    flag: Japan                  jp         Flags                 1
#> 2 😀    grinning face                grinning   Smileys & Emotion     1
#> 3 😍    smiling face with heart-eyes heart_eyes Smileys & Emotion     1
#> 4 😡    enraged face                 rage       Smileys & Emotion     1
#> 5 🚀    rocket                       rocket     Travel & Places       1
emoji_sentiment(reviews, text)
#> # A tibble: 4 × 4
#>   text                         .emoji_n .emoji_n_scored .emoji_sentiment
#>   <chr>                           <int>           <int>            <dbl>
#> 1 Best purchase ever 😀😍             2               2            0.625
#> 2 It broke after a day 😡             1               1           -0.173
#> 3 Does the job.                       0              NA           NA    
#> 4 Shipped to 🇯🇵 in two days 🚀        2               1            0.525

How much of that could the lexicon score? One more call says so:

emoji_coverage(reviews, text)[, c("lexicon", "occurrence_coverage",
                                  "n_rows_unscored")]
#> # A tibble: 2 × 3
#>   lexicon    occurrence_coverage n_rows_unscored
#>   <chr>                    <dbl>           <int>
#> 1 novak2015                  0.8               0
#> 2 emotag1200                 0.6               1

🚀 Get started

Install from CRAN: install.packages("tidyEmoji")
Load it: library(tidyEmoji)
Hand any verb a data frame and its text column: emoji_summary(df, text)
Take the tour: vignette("introduction", package = "tidyEmoji")

🧰 Pick a verb by question

Question	Verbs
How much emoji is here, and which?	`emoji_summary()`, `emoji_frequency()`, `emoji_tokens()`
What does it say?	`emoji_sentiment()`, `emoji_emotion()`, `emoji_coverage()`
Will readers agree on it?	`emoji_ambiguity()`, `emoji_risk()`
What is this glyph, exactly?	`emoji_properties()`, `emoji_identical()`, `emoji_find()`
What goes with what?	`emoji_pairs()`, `emoji_collocations()`, `emoji_ngrams()`
When, and how new?	`emoji_trend()`, `emoji_adoption_lag()`
Which countries do the flags name?	`emoji_country()`, `emoji_flag()`
Reaction counts rather than text?	`emoji_from_counts()`
Sending text to a language model?	`emoji_sanitize()`, `emoji_token_cost()`

💡 Tips

Count bare text-style emoji such as the plain heart with presentation = "any", which every counting and scoring verb takes.
Report emoji_coverage() next to any lexicon score: it says how much of the corpus the score covers.
Paste emoji_provenance() into a methods section: it records the Unicode catalogue (Emoji 18.0, bundled) and the lexicons behind a result.

📚 Learn more

Three articles ship with the package: vignette("introduction") for a full analysis of a real corpus, vignette("reversible-preprocessing") for emoji in language-model pipelines, and vignette("emoji-networks") for co-occurrence networks. All three are on the package website.