Real-time expressive internet communications.

Abstract

This research work "Real-time Expressive Internet Communications" focuses on two subjects: One is the investigation of methods of automatic emotion detection and visualisation under real-time Internet communication environment, the other is the analysis of the influences of presenting visualised emotion expressivei mages to Internet users. To detect emotion within Internet communication, the emotion communication process over the Internet needs to be examined. An emotion momentum theory was developed to illustrate the emotion communication process over the Internet communication. It is argued in this theory that an Internet user is within a certain emotion state, the emotion state is changeable by internal and external stimulus (e.g. a received chat message) and time; stimulus duration and stimulus intensity are the major factors influencing the emotion state. The emotion momentum theory divides the emotions expressed in Internet communication into three dimensions: emotion category, intensity and duration. The emotion momentum theory was implemented within a prototype emotion extraction engine. The emotion extraction engine can analyse input text in an Internet chat environment, detect and extract the emotion being communicated, and deliver the parameters to invoke an appropriate expressive image on screen to the every communicating user's display. A set of experiments were carried out to test the speed and the accuracy of the emotion extraction engine. The results of the experiments demonstrated an acceptable performance of the emotion extraction engine. The next step of this study was to design and implement an expressive image generator that generates expressive images from a single neutral facial image. Generated facial images are classified into six categories, and for each category, three different intensities were achieved. Users need to define only six control points and three control shapes to synthesise all the expressive images and a set of experiments were carried out to test the quality of the synthesised images. The experiment results demonstrated an acceptable recognition rate of the generated facial expression images. With the emotion extraction engine and the expressive image generator,a test platform was created to evaluate the influences of emotion visualisation in the Internet communication context. The results of a series of experiments demonstratedthat emotion visualisation can enhancethe users' perceived performance and their satisfaction with the interfaces. The contributions to knowledge fall into four main areas; firstly, the emotion momentum theory that is proposed to illustrate the emotion communication process over the Internet; secondly, the innovations built into an emotion extraction engine, which senses emotional feelings from textual messages input by Internet users; thirdly, the innovations built into the expressive image generator, which synthesises facial expressions using a fast approach with a user friendly interface; and fourthly, the identification of the influence that the visualisation of emotion has on human computer interaction.