Group items tagged

Having learned a good representation for emoji, we can begin to ask questions about similarity. Namely, for a given emoji, what English words are semantically similar? For each emoji, we compute the “angle” (equivalently the cosine similarity) between it and other words. Words with a small angle are said to be similar and provide a natural, English-language translation for that emoji.

Using our algorithm, we find that many of our popular emoji have meanings in-line with early internet slang:

It seems that the most popular emoji have similar semantics to words like “lol/hehe” (

Many clusters emerge: food emoji on the left, opposite the work emoji in the top right. Shoes (bottom right) are associated closely to handbags while bathing suits are closer to the water and marine animals (top left). Alcoholic drinks (bottom left) cluster together with bowling. Towards the center, we see a large clustering of facial expressions bordered by sadness, shock, laughter, happiness and coolness. As we travel downwards, we can see happy, love leading all the way the family and wedding emoji.

On Instagram, emoji are becoming a valid and near-universal method of expression in all languages. Emoji usage is shifting the people’s vocabulary on Instagram and becoming an important means of expression: their use is anti-correlated with internet slang like “lol” and “xoxo.”

when texters finish a conversation, they often trade a few emoji as nonverbal denouement. “You might not have anything else left to say,” Kelly says, “but you want to let the person know that you’re thinking of them.”

people are even developing syntax and rules of use for emoji. Schnoebelen found that when we use face emoji, we tend to put them before other objects. If you text about a late flight, you’ll put an unhappy face followed by a plane, not the reverse. In linguistic terms, this is called conveying “stance.” Just as with in-person talk, the expression illustrates our stance before we’ve spoken a word.

Churches wanted to find out if the same applied when we looked at a smiley face emoticon, which is a stylised representation of a smiling human face.

The smiley face emoticon first appeared in a post to Carnegie Mellon University computer science general board from Professor Scott E Fahlman in 1982.

Since then, the same pattern of activity as evoked by faces has become attached to what was previously just punctuation.

you can provide the influencer with some pointers. Would you prefer they include keywords in the title of their product review blog to boost your SEO? How many hashtags do you want them to use, and are fans likelier to adopt shorter ones over longer ones? Should they use emojis? (The answer is almost always yes.) Which CTA performs better, “Save 15% off now with my promo code” or “Use my promo code now”?

Speaking of promo codes, what learnings can you apply from sales you’ve run in the past? Does a percentage or dollar off amount drive more conversions? Does what works for sales on your own website work just as well in the context of an influencer promotion?

Perhaps influencers’ fans are more excited about getting a free sample or trial instead of a discount. In this scenario, try testing free sample promotions with some influencers against discount offers with other influencers. Just be sure to choose influencers with similar audiences, industries, and/or locations to keep the other variables as similar as possible.

A/B test the heck out of your influencer landing pages. Try different CTA button placements and colors, test removing the navigation, and see how personalizing the page for the influencer’s audience affects conversions.

Not all your influencer marketing content is published by the influencer. Sometimes, as with the landing pages, you are using the influencers in your own content. A/B test the items under your branded control, too.

if you feature an influencer in an email newsletter, is it best to call that out in the subject line, via the sender name, through a hero image at the top, or some combination of the above? Should you target different subscriber lists for different featured influencers