未來電腦可以僅通過Emoji表情就能識別諷刺?

Emoji icons have helped scientists train computers to understand sarcasm.

Emoji表情符號有助於科學家訓練電腦以理解諷刺。

Researchers used 1.2 billion tweets containing at least one of the most popular 64 emojis to develop DeepMoji.

研究人員收集了12億條推文，每條推文都至少有64種最受歡迎表情符號中的一個，這是爲了開發DeepMoji。

The algorithm learned first to predict which emoji was likely to be used and then to recognise sarcasm, allowing it to spot hate speech faster than humans.

該算法首先學會了預測哪些表情可能被使用，然後識別言語中的諷刺情緒，讓它能比人類更快地發現仇恨言論。

It could also be used to assess how people really feel about different brands or to improve human interaction with smart programs such as chatbots.

它還可以用來評估人們對不同品牌的真實感受，或者改進與智能程序(如聊天機器人)的人際互動。

"Because we can't use intonation in our voice or body language to contextualise what we are saying, emojis are the way we do it online," Prof Iyad Rahwan told MIT's Technology Review magazine.

Iyad Rahwan向麻省理工學院的技術評論雜誌透露：“因爲在網絡上，我們不會使用自己的聲音或肢體語言來將我們所說的內容語境化，而是用表情符號。”

Prof Iyad Rahwan and graduate student Bjarke Felbo have created a website inviting people to refine DeepMoji's education by submitting and annotating their own messages.

Iyad Rahwan教授和研究生Bjarke Felbo創建了一個網站，邀請人們通過提交和註釋自己的信息（中的情緒）來改進DeepMoji的訓練。

The pair also plan to release the algorithm's code so it can be used by other researchers.

他們還計劃公佈算法的代碼，以便其他研究人員使用。