Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers

Jeroen Dral; Dirk K.J. Heylen; Hendrikus J.A. op den Akker

doi:10.1007/978-94-007-1757-2_6

Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers

Jeroen Dral, Dirk K.J. Heylen, Hendrikus J.A. op den Akker

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

22 Downloads (Pure)

Abstract

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.

Original language	Undefined
Title of host publication	Affective Computing and Sentiment Analysis
Editors	Kurshid Ahmad
Place of Publication	London
Publisher	Springer
Pages	67-77
Number of pages	11
ISBN (Print)	978-94-007-1756-5
DOIs	https://doi.org/10.1007/978-94-007-1757-2_6
Publication status	Published - 2011

Publication series

Name	Text, Speech, and Language Analysis
Publisher	Springer Verlag
Number	45
Volume	45
ISSN (Print)	1386-291X

Keywords

METIS-285103
IR-79612
EWI-21454
sentiment analysis prosody uncertainty
HMI-MR: MULTIMEDIA RETRIEVAL
EC Grant Agreement nr.: FP6/033812

Access to Document

10.1007/978-94-007-1757-2_6

http://eprints.eemcs.utwente.nl/secure2/21454/01/detectinguncertainty.pdf

Cite this

Dral, J., Heylen, D. K. J., & op den Akker, H. J. A. (2011). Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers. In K. Ahmad (Ed.), Affective Computing and Sentiment Analysis (pp. 67-77). (Text, Speech, and Language Analysis; Vol. 45, No. 45). Springer. https://doi.org/10.1007/978-94-007-1757-2_6

@inbook{817244dbf06347918adbeedc3b4598eb,

title = "Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers",

abstract = "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.",

keywords = "METIS-285103, IR-79612, EWI-21454, sentiment analysis prosody uncertainty, HMI-MR: MULTIMEDIA RETRIEVAL, EC Grant Agreement nr.: FP6/033812",

author = "Jeroen Dral and Heylen, {Dirk K.J.} and {op den Akker}, {Hendrikus J.A.}",

note = "10.1007/978-94-007-1757-2_6 ",

year = "2011",

doi = "10.1007/978-94-007-1757-2_6",

language = "Undefined",

isbn = "978-94-007-1756-5",

series = "Text, Speech, and Language Analysis",

publisher = "Springer",

number = "45",

pages = "67--77",

editor = "Kurshid Ahmad",

booktitle = "Affective Computing and Sentiment Analysis",

address = "Germany",

}

Dral, J, Heylen, DKJ & op den Akker, HJA 2011, Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers. in K Ahmad (ed.), Affective Computing and Sentiment Analysis. Text, Speech, and Language Analysis, no. 45, vol. 45, Springer, London, pp. 67-77. https://doi.org/10.1007/978-94-007-1757-2_6

Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers. / Dral, Jeroen; Heylen, Dirk K.J.; op den Akker, Hendrikus J.A.
Affective Computing and Sentiment Analysis. ed. / Kurshid Ahmad. London: Springer, 2011. p. 67-77 (Text, Speech, and Language Analysis; Vol. 45, No. 45).

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

TY - CHAP

T1 - Detecting Uncertainty in Spoken Dialogues: An explorative research for the automatic detection of speaker uncertainty by using prosodic markers

AU - Dral, Jeroen

AU - Heylen, Dirk K.J.

AU - op den Akker, Hendrikus J.A.

N1 - 10.1007/978-94-007-1757-2_6

PY - 2011

Y1 - 2011

N2 - 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.

AB - 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.

KW - METIS-285103

KW - IR-79612

KW - EWI-21454

KW - sentiment analysis prosody uncertainty

KW - HMI-MR: MULTIMEDIA RETRIEVAL

KW - EC Grant Agreement nr.: FP6/033812

U2 - 10.1007/978-94-007-1757-2_6

DO - 10.1007/978-94-007-1757-2_6

M3 - Chapter

SN - 978-94-007-1756-5

T3 - Text, Speech, and Language Analysis

SP - 67

EP - 77

BT - Affective Computing and Sentiment Analysis

A2 - Ahmad, Kurshid

PB - Springer

CY - London

ER -