Abstract

This paper describes the latest developments of a generic prototype a real-time text-to-emotion analyser. The core component of the analyser is the emotion extraction engine. The engine can analyse input text, extract the contained emotion and deliver the parameters necessary to invoke an appropriate image expressing the emotion. The parameters include the interpreted emotion and intensity, both represented by the displayed image. A set of rules are defined to control the behaviour of the emotion extraction engine. The concept of ‘Emotional Momentum’ is implemented in the average weighted emotional mood indicator to deal with ambiguity or conflicting emotional content. The engine has been applied in a number of different environments including real-time expressive communication systems over the Internet and journal article analysis systems. We present experimental results carried out to test the engine’s performance. The results illustrate that the analyser can be used successfully within the designed environments.