In recent years there has been a growth of research focusing on speech-based interactive systems that aim to increase the system’s communicative competence by including aspects of interaction beyond simple speech recognition and question-answer based interaction. Such advanced aspects include speech segmentation, disfluencies, turn-taking, emotions, and adaptation, and these aspects have been studied from the view-point of both practical applications and theoretical interaction modelling.

However, it is often felt that statistical speech-based research is not fully appreciated in the dialogue community, while dialogue modelling in the speech community seems too simple in terms of the advanced architectures and functionalities under investigation in the dialogue community. One of the motivations for furthering the system’s interaction capabilities on both sides is to improve the system’s naturalness and usability in practical applications. Relatively little work has so far been devoted to defining the criteria according to which we could evaluate speech-based interaction systems in these terms. From the industrial point of view, academically developed research is often too far removed from what is needed in practical applications, while academic researchers often feel that their research results do not find their way into industrial applications.

The goal of this one-day workshop is to encourage dialogue on dialogues: multidisciplinary discussions on the evaluation criteria for speech-based systems, descriptions of completed or on-going research, and debates on what is still needed. The workshop aims to bring together researchers and practitioners working on the development of communication models that support robust and efficient interaction in spoken natural language both for commercial dialogue systems and in basic research and, through discussions and working sessions, identify and enumerate criteria for evaluating robust and efficient interaction in spoken dialogue systems.

Organizers: