Emoji Sentiment

Emoji sentiment data

• lightweight, easy to use JSON data with a minimalistic API
• based on the work of Kralj Novak, Petra; Smailović, Jasmina; Sluban, Borut and Mozetič, Igor, 2015, Emoji Sentiment Ranking 1.0, Slovenian language resource repository CLARIN.SI, http://hdl.handle.net/11356/1048.
• they engaged 83 human annotators to label over 1.6 million tweets in 13 european languages by sentiment polarity (negative, neutral or positive) while about 4% of the annotated tweets contained emoji
• this module provides transformed JSON data based on the original CSV data and derived data calculated in compliance with the original paper

Have a look at this table to see an example of what data this library provides.

API

Requiring/importing emoji-sentiment gives you the following data structure to work with:

[..., { /* emoji sentiment datum */ }, ...]

Array of emoji sentiment data.

Example of an emoji sentiment datum:

[
  ...
  {
    // original properties:
    "sequence": "1F602",
    "occurrences": 14622,
    "negative": 3614,
    "neutral": 4163,
    "positive": 6845,
    // derived properties:
    "pNegative": 0.24717948717948718,
    "pNeutral": 0.2847179487179487,
    "pPositive": 0.4681025641025641,
    "score": 0.22092307692307694,
    "sem": 0.006751317877016391
  },
  ...
]

Original properties are included in the JSON file whereas derived properties are being calculated on demand when the module is being required/imported.

Properties of an emoji sentiment datum explained:

• sequence (original)

  normalized code point sequence (sequence without any variation selector or modifier applied) e.g. 1F602; use it for mapping the sentiment datum to a specific (emoji) unicode character or connecting it with further meta data (e.g. unicode-emoji-data, unicode-emoji-annotations or emoji-datasource)

• occurrences (original)

  absolute number of occurrences of the (emoji) unicode character in tweets

• negative (original)

  absolute number of occurrences of the (emoji) unicode character in tweets labeled negative

• neutral (original)

  absolute number of occurrences of the (emoji) unicode character in tweets labeled neutral

• positive (original)

  absolute number of occurrences of the (emoji) unicode character in tweets labeled positive

• pNegative (derived)

  relative negativity component of the sentiment distribution for those tweets associated with the (emoji) unicode character, ranging from 0 to 1

• pNeutral (derived)

  relative neutrality component of the sentiment distribution for those tweets associated with the (emoji) unicode character, ranging from 0 to 1

• pPositive (derived)

  relative positivity component of the sentiment distribution for those tweets associated with the (emoji) unicode character, ranging from 0 to 1

• score (derived)

  resulting sentiment score of the (emoji) unicode character, ranging from -1 to +1, calculated as the mean of the discrete sentiment distribution of negative (-1), neutral (0) and positive (+1)

• sem (derived)

  precalculated Standard Error Mean for further deriving the confidence interval, e.g. for 95%: [score − 1.96 * sem, score + 1.96 * sem]

The sum of negative, neutral and positive is occurrences.

The sum of pNegative, pNeutral and pPositive is 1.

Usage

CommonJS

const emojiSentiment = require('emoji-sentiment');

ES6/babel

import emojiSentiment from 'emoji-sentiment';

Install

npm install emoji-sentiment

License

MIT

Development

Status