Dear Eiman Kanjo, 

Look the link. May be useful.

http://people.few.eur.nl/frasincar/papers/SAC2013b/sac2013b.pdf

https://arxiv.org/ftp/arxiv/papers/1511/1511.02556.pdf

Regards, Shafagat

Thank you Shafagat, I have already read these two papers. They looked at Emoicon rather than emojies also their work is based on text from the web while I am interested in social media feed (whatsup, viber..etc).

Please check out this link.

https://scholar.google.co.in/scholar?as_ylo=2015&q=algorithm+for+sentiment+analysis+on+emojies%3F&hl=en&as_sdt=0,5

Thanks and regards,

Rekha Mishra

Emoji's on their own lack sentiment that comes from the context of the conversation. Any algorithm dealing with this should look at who the person is, and what was said which are what add meaning to pictures beyond what first glance would suggest.

For example the smiley face emoticon :-) was originally developed to show sarcasm and light heartedness.

Online tools like http://www.thesarcasmdetector.com are basic examples at understanding the nuiance of language. "I like Nickelback" scores -13 meaning less likely to be sarcastic than "I like Nickelback :D!" which scores +8.

References and info for this are provided in an about page.

In addition meaning can even vary from the device it is sent from (e.g. fat dancer android vs sexy red dress iOS)

http://emojipedia.org/dancer/

but at the end of the day. I do like Nickelback :-)