Using past speaker behavior to better predict turn transitions

Tomer Meshorer, Peter Heeman

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

This paper explores using a summary of past speaker behavior to better predict turn transitions. We computed two types of summary features that represent the current speaker's past turn-taking behavior: relative turn length and relative floor control. Relative turn length measures the current turn length so far (in time and words) relative to the speaker's average turn length. Relative floor control measures the speaker's control of the conversation floor (in time and words) relative to the total conversation length. The features are recomputed for each dialog act based on past turns of the speaker within the current conversation. Using the switchboard corpus, we trained two models to predict turn transitions: one with just local features (e.g., current speech act, previous speech act) and one that added the summary features. Our results shows that using the summary features improve turn transitions prediction.

Original languageEnglish (US)
Pages (from-to)2900-2904
Number of pages5
JournalUnknown Journal
Volume08-12-September-2016
DOIs
StatePublished - 2016

Fingerprint

Predict
Local Features
Length
Summary
Prediction
Speech Acts
Speech
Model

Keywords

  • Conversation
  • Speaker transition
  • Turn taking

ASJC Scopus subject areas

  • Language and Linguistics
  • Human-Computer Interaction
  • Signal Processing
  • Software
  • Modeling and Simulation

Cite this

Using past speaker behavior to better predict turn transitions. / Meshorer, Tomer; Heeman, Peter.

In: Unknown Journal, Vol. 08-12-September-2016, 2016, p. 2900-2904.

Research output: Contribution to journalArticle

Meshorer, Tomer ; Heeman, Peter. / Using past speaker behavior to better predict turn transitions. In: Unknown Journal. 2016 ; Vol. 08-12-September-2016. pp. 2900-2904.
@article{3a62690d39fa41319ae6f22697937be8,
title = "Using past speaker behavior to better predict turn transitions",
abstract = "This paper explores using a summary of past speaker behavior to better predict turn transitions. We computed two types of summary features that represent the current speaker's past turn-taking behavior: relative turn length and relative floor control. Relative turn length measures the current turn length so far (in time and words) relative to the speaker's average turn length. Relative floor control measures the speaker's control of the conversation floor (in time and words) relative to the total conversation length. The features are recomputed for each dialog act based on past turns of the speaker within the current conversation. Using the switchboard corpus, we trained two models to predict turn transitions: one with just local features (e.g., current speech act, previous speech act) and one that added the summary features. Our results shows that using the summary features improve turn transitions prediction.",
keywords = "Conversation, Speaker transition, Turn taking",
author = "Tomer Meshorer and Peter Heeman",
year = "2016",
doi = "10.21437/Interspeech.2016-1409",
language = "English (US)",
volume = "08-12-September-2016",
pages = "2900--2904",
journal = "Indian Journal of Rheumatology",
issn = "0973-3698",
publisher = "Elsevier (Singapore) Pte Ltd",

}

TY - JOUR

T1 - Using past speaker behavior to better predict turn transitions

AU - Meshorer, Tomer

AU - Heeman, Peter

PY - 2016

Y1 - 2016

N2 - This paper explores using a summary of past speaker behavior to better predict turn transitions. We computed two types of summary features that represent the current speaker's past turn-taking behavior: relative turn length and relative floor control. Relative turn length measures the current turn length so far (in time and words) relative to the speaker's average turn length. Relative floor control measures the speaker's control of the conversation floor (in time and words) relative to the total conversation length. The features are recomputed for each dialog act based on past turns of the speaker within the current conversation. Using the switchboard corpus, we trained two models to predict turn transitions: one with just local features (e.g., current speech act, previous speech act) and one that added the summary features. Our results shows that using the summary features improve turn transitions prediction.

AB - This paper explores using a summary of past speaker behavior to better predict turn transitions. We computed two types of summary features that represent the current speaker's past turn-taking behavior: relative turn length and relative floor control. Relative turn length measures the current turn length so far (in time and words) relative to the speaker's average turn length. Relative floor control measures the speaker's control of the conversation floor (in time and words) relative to the total conversation length. The features are recomputed for each dialog act based on past turns of the speaker within the current conversation. Using the switchboard corpus, we trained two models to predict turn transitions: one with just local features (e.g., current speech act, previous speech act) and one that added the summary features. Our results shows that using the summary features improve turn transitions prediction.

KW - Conversation

KW - Speaker transition

KW - Turn taking

UR - http://www.scopus.com/inward/record.url?scp=84994301844&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84994301844&partnerID=8YFLogxK

U2 - 10.21437/Interspeech.2016-1409

DO - 10.21437/Interspeech.2016-1409

M3 - Article

AN - SCOPUS:84994301844

VL - 08-12-September-2016

SP - 2900

EP - 2904

JO - Indian Journal of Rheumatology

JF - Indian Journal of Rheumatology

SN - 0973-3698

ER -