Workshop Proceedings of the 14th International AAAI Conference on Web and Social Media
To date, a significant body of literature has examined the unique characteristics involved in communicating emotion. These days, emotional intent is frequently conveyed through emoji in online communications. If we are to design technology that assists in conveying emotional intent, we need to better understand how particular design details can play a role in interpretation - particularly across different vendors, many of which have slightly varying designs of the same emoji. With this research we compared the sentiment associated with 45 emoji across four commonly used vendors (iOS, Android, Samsung, and WhatsApp) to identify areas of misalignment and begin identifying features that influence sentiment interpretation. 6,250 US mobile phone users were asked to select the sentiment that best described 10 randomly-selected emoji. Results indicate a high level of sentiment agreement for emoji expressing positive affect, surprise, and those with external cues (such as a Face with Medical Mask). Lower levels of agreement were assigned to emoji that expressed negative affect, ambiguous expressions, or had closed eyes. Potential reasons for the low levels of agreement include less familiarity with specific emoji and differences between vendors in how they illustrate facial features.