Emoji, the smileys and icons representing an idea or emotion and originating in Japan over a decade ago, are becoming increasingly popular on social networks. According to Instagram’s research, this popularity rose with support for emoji being added to major mobile operating systems iOS and Android. I’ve seen emoji used so frequently on social networks this year that in 2016 we should see a growth in analytics focused specifically on emoji.
Indeed, VentureBeat confirms that emoji have a role to play in social analytics trends in 2016. But what kind of meaningful results can be extracted from the use of emoji? I can think of two kinds:
Improving the accuracy of sentiment analysis
Through the use of natural language processing and the ranking of sentences by their assumed meaning, social posts can be sorted by emotion, or by determining whether a message is expressing a positive or negative opinion about a topic. This can be a complex process, relying on machine learning to understand the meaning of words with multiple meanings, slang, and abbreviations within a sentence.
When relying solely on text, there’s also risk of misinterpreting a comment due to hard-to-detect sarcasm and irony. However, facial expressions, emotions, and to a lesser extent well-known symbols such as a heart ❤ are understood by many, and have interpretations that stand the test of time. Since these are so universal, smiley face-type emoticons and symbol emojis are perfectly convenient for boosting sentiment recognition.
Consider two comments: one which is text-only, and another with the same text but also ends with a smile emoji. Sentiment analysis of the first comment may indicate that the comment is a positive one, but with low level of certainty. But in the second example, the addition of a smiley-face emoji to that comment would give the algorithm increased confidence that the initial analysis was correct.
Of course, users can still misuse (or abuse) emoji. Just as sarcasm complicates sentiment analysis, using emoji in a comment that’s in apparent contradiction to the text will certainly complicate matters. Regardless, analysis of emoji with a comment can be deployed as another useful tool to boost certainty of an analysis.
Extracting additional meaning from emoji messages through emoji-word relationships
This is a bit harder to understand intuitively compared to sentiment analysis. Generally, there are no standard conventions that dictate when or where emoji are used in a sentence, but occasionally you’ll find social comments in which emoji seem to have been used in place of a word. Or instead, you may find emoji at the end of a sentence, but you many not understand the intended usage. Instagram’s engineering blog recently published a great piece on combining machine learning and emoji to discover relationships between emoji and phrases. In the article, some interesting relationships are discovered, and I think there is additional insight to be gained from comment analysis when the many possible meaning of emoji are considered.
It’s hard to say if emoji are here to stay or simply a brief fad as other forms of Internet slang have been. However, while they last, emoji are not just a way to give a comment more personality or keep a twitter post under the 140 character limit, but potentially a handy tool for increasing the accuracy of data coming from social analytics tools.
Latest posts by Sean Grady (see all)
- How can emoji analysis improve results in social analytics and social listening tools? - May 13, 2016
- Simplified storage configuration on the EMC VPLEX - April 8, 2016
- A Quick Look at VPLEX Storage Elements - March 11, 2016