Communicating Emotions to Virtual Agents: An Emoticon-Based Approach

1st Supervisor: Prof. Dr. Bernhard Jung (ISNM)

2nd Supervisor: Prof. Dr. Andreas Schrader (ISNM)

With the development of artificial intelligence technologies, there has arrived a possibility to have natural language conversation with three-dimensional, real-time animated virtual agents. Text based natural language conversation, however, suffers from lack of non-verbal communication channels. The capability of virtual agents to understand emotions in natural language conversation would enhance their believability. In other work, face recognition techniques have been tried to convey emotions from humans to virtual agents. This research focuses on using emoticons to communicate emotions in human-computer interaction. The use of emoticons has an advantage that it requires no external hardware devices. Emoticons can be used as a added communication channel augmenting natural language input in human computer interaction. A prototype has been developed which consists of a virtual character with natural conversation and appropriate facial expression abilities. The natural language understanding capabilites were extended to allow for the use of emotions when talking to the agent. A group of users was asked to interact with the character and asked to submit a feedback pertaining to certain questions. The analysis of the feedback suggests that conversational agents having emotions in them will significantly improve the interaction and believability.