MMSDS: Ubiquitous computing and WWW-based multi-modal sentential dialog system

Abstract

In this study, we suggest and implement Multi-Modal Sentential Dialog System (MMSDS) integrating 2 sensory channels with speech and haptic information based on ubiquitous computing and WWW for clear communication. The importance and necessity of MMSDS for HCI as following: 1) it can allow more interactive and natural communication functions between the hearing-impaired and hearing person without special learning and education, 2) according as it recognizes a sentential Korean Standard Sign Language (KSSL) which is represented with speech and haptics and then translates recognition results into a synthetic speech and visual illustration in real-time, it may provide a wider range of personalized and differentiated information more effectively to them, and 3) above all things, a user need not be constrained by the limitations of a particular interaction mode at any given moment because it can guarantee mobility of WPS (Wearable Personal Station for the post PC) with a built-in sentential sign language recognizer. In experiment results, while an average recognition rate of uni-modal recognizer using KSSL only is 93.1% and speech only is 95.5%, advanced MMSDS deduced an average recognition rate of 96.1% for 32 sentential KSSL recognition models. © IFIP International Federation for Information Processing 2006.