Utilizing Statistical Dialogue Act Processing in Verbmobil

TL;DR

This paper introduces a statistical method for dialogue act processing in a speech translation system, enhancing its ability to predict dialogue flow and manage unexpected states through repair mechanisms.

Contribution

It presents a novel statistical approach for dialogue act prediction and repair support in speech-to-speech translation systems.

Findings

Improved prediction of follow-up dialogue acts.

Enhanced repair capabilities during unexpected dialogue states.

Demonstrated effectiveness within the VM speech translation system.

Abstract

In this paper, we present a statistical approach for dialogue act processing in the dialogue component of the speech-to-speech translation system \vm. Statistics in dialogue processing is used to predict follow-up dialogue acts. As an application example we show how it supports repair when unexpected dialogue states occur.