Detecting Uncertainty in Spoken Dialogues An explorative research to the automatic detection of a speakers' uncertainty by using prosodic markers

This paper reports results in automatic detection of speaker uncertainty in spoken dialogues by using prosodic markers. For this purpose a substantial part of the AMI corpus (a multi-modal multi-party meeting corpus) has been selected and converted to a suitable format so its data could be analyzed for a selected set of prosodic features. In the absence of relevant stance annotations on (un)certainty, lexical markers (hedges) have been used to mark utterances as (un)certain. Results show that prosodic features can indeed be used to detect speaker uncertainty in spoken dialogues. The classifiers can tell uncertain from neutral utterances with an accuracy of 75% which is 25% over the baseline.