Expressing emotions as emoticons for online intelligent agents

Without emotional annotation, online communication can be ambiguous and lead to misunderstandings. This paper addresses the questions of which emotions are commonly expressed online, how these emotions can be encapsulated in emoticons, and how people respond to different emotions. In 10 focus groups with university students we found that some emotions are not frequently expressed online (e.g. aggravation, alienantion and torment), while many others were commonly used (e.g. enthusiasm, anger, amusement, amazement and disgust). Emoticons were drawn or described for nine commonly expressed emotions, and the response discussed. Audience was a key component in how people used emoticons, both for use and interpretation. Participants preferred to ‘defuse’ negative emotions such as anger and rage with light-hearted comments, supporting previous findings on a positivity bias on many social networks. These findings have implications for online communication and the design of intelligent virtual agents.